What is on your workbench right now?

[restorer-john]

The second of my little PA5 IIs undergoing Neutrifikation, ready for the 2-way desktop speakers', active conversion:

View attachment 429160

View attachment 429161

View attachment 429162
Yeah, one of them's upside down. Had to unscrew it and turn it around

View attachment 429163

Great job.

Call me old fashioned, but I want an amplifier big and heavy enough that the plugs and cables can't pull it off the table...

 

[egellings]

Great job.

Call me old fashioned, but I want an amplifier big and heavy enough that the plugs and cables can't pull it off the table...

You'd like my home-brewed vacuum tube mono blocks. 85 pounds apiece, and you need one for each ear. I have them sited on the floor.

 

[gfinlays]

Great job.

Call me old fashioned, but I want an amplifier big and heavy enough that the plugs and cables can't pull it off the table...

Thanks!

All of mine are Class D these days, though the 6 channels of Hypex in the lounge are a bit more substantial!

The PA5 IIs do the job for the desktop setup, and they don't take up much space but it would be nice if they were a bit heavier. Consequence of an external PSU. Still, at least the speaker cables won't be falling out anytime soon

 

[AnalogSteph]

Yamaha Pianocraft RX-E410 mini receiver, fished out of the bay motivated by the recent success with an RX-E810 in the local repair café (wouldn't stay on - C8 measured sub-10 nF, a new 22n Y2 did the trick). As advertised, it's quite dead. Fairly good shape optically, not a ton of dust, some musty basement smell.

The notorious 2000s Yamaha standby power supply (on which I think I'll be writing a theory of operation soon - EDIT: done) isn't well-protected, and by the list of casualties I'd say this one suffered a major surge while turned off.
Capacitor dropper C8 (22n 630V polyester) - dead short! (that should really have been a Y class type)
Uh-oh...
R161 (it's a tiny through-hole, not 4x SMD R161-164 as shown in service docs), D9, D10, D8, R11, all blown open and more or less toasty. J32 looks like it got warm.
Optocoupler IC2 diode, short. Dead short in parallel to C9/IC3 (probably the classic of the semiconductor protecting the capacitor). Q4 D-S or D11, dead short.

In less bad news, the board looks rebuildable, both fuses are OK, the standby transformer gives me 3.66 kOhms on the primary and the main transformer 10.69 ohms (which seems kinda low but not impossibly so). Nothing obviously bad on the secondary side so far.

So much for just dropping in a new C8 and having a nice little receiver for my parents' PC sound that doesn't hiss... (EDIT: Got them a slightly less cosmetically nice RX-E400 with an all-conventional power supply instead. That works fine, the nodes between volume pot wipers and preamp inputs just seem to have slightly meh shielding as there is some buzz at higher volumes that peaks at -6 dB and goes away at max, so clearly something related to source impedance. Otherwise it's quite "proper" hi-fi and definitely an improvement on the SA-PM01. Power consumption: standby 0.65 W, idle starting at ~13.2 W and quickly rising to ~13.65 W, creeping up to ~13.9 W over several hours, and even playing fairly loud does not generally add more than 1-2 W. For comparison, the SA-PM01 was at ca. 4 W standby and 8 W idle.)

My plan for testing the rest once I have access to the equipment required:
Unsolder whatever is shorting Q4 || D11
Supply standby secondary with 10.2 V from lab PSU
Use toggle switch across IC2 transistor to make device think it does / doesn't have mains power.

Results will determine the further course of action. If it turns out to be a braindead toasted brick this one will only be for spares.

Preliminary thoughts on circuit upgrades:
C8 will be a Y2, that's a given. The pin spacing is actually too small for the cap that's in there already, that was an issue with the 810 as well. Might just recycle part of the old cap's legs.
R161 needs to be higher if it's supposed to protect anything. Vector addition suggests something like 33k should not unduly impact |Z| of R+C yet. Ideally it would be a carbon comp, but maybe two MFs in series?
D9 got blown open completely, so I wouldn't mind putting in something beefier than those little SOD323 low-noise jobs. Of course a number of parts are stupid tiny SMD crap bunched together.

 

[JP]

Helping some folks bring some Technics 'tables back to life.

 

[RickS]

New crossover for Sica coaxial…

 

[AnalogSteph]

My trusty Kenwood KT-80 bedside FM tuner from 1980. I finally wanted to sort out the lighting situation, which despite the power indicator bulb already having failed is very inefficient... that poor dropper resistor is getting veeery toasty. (15.95 Vrms over 180 ohms, you do the math.) I'm actually measuring 65 ohms in both directions, so I think the poor 14.2 V zener has gone severely leaky over the years. The current plan is to replace the toasty resistor with a 3.3µ/100V film cap, install an additional rectifier diode to turn the whole shebang into a Greinacher circuit (normally a voltage doubler), and swap the half-dead zener with a small 8.2 V part, so I can have a nice stable 8 V supply with fairly low losses. I've seen a 5 mm diode for 12 V with built-in dropper which may eventually replace the dead bulb.


While in there, I measured a few things for the hell of it and noticed relatively poor load regulation of the main +14 V supply, which is actually almost 15 V. Between "no station, no LEDs lit" and "all 5 S-meter LEDs lit, stereo, AFC" I saw about a 92 mV delta. Considering they even went to the effort of doing remote voltage sensing (this should ring a bell with anyone familiar with the Kenwood power amps of the time and their "Sigma Drive"), that struck me as rather mediocre.

Turns out that Kenwood duly RC-filtered the zener reference (in itself rock-stable at about 6.078 V), which makes the effective reference voltage slightly dependent upon unregulated input voltage. And that in turn varies a fair bit with mains transformer loading. I've seen as high as 22.1 V and as low as 19.95 V. If I remove the toastiness in the lighting circuit, that's not going to get any better.

I thought of using a feedforward bodge resistor and eventually came up with this, which as projected will take two resistor and one capacitor (I used transient and AC simulation to optimize values):

You can see how I(R117) is basically equal to I(R115 + R116), and the voltage across those two equals V+ - (Vreg + 2Vbe), which in turn varies between 3.74 and 5.88 V according to my measurements. I've seen between 59 and 94 mV across R117. Rff1 + Rff2 slightly tug on the feedback node to offset the minor reference voltage change, with Cff making sure we're not degrading our AC ripple rejection. It doesn't work perfectly but you can get a 20-30 dB improvement over a certain range. (In sim, ~160 mV went down to the mere 4 mV spread seen above.) Not bad for 3 passives!

(My first attempt using a transistor CCS instead of feedforward was a bit of a bust, but I have an idea what went wrong, so I'll try that again as well.)

Projected DC output using the actual zener voltage (I had to resort to a 6.2 V part for simulation) is going to be 14.24 V. If I actually get a sub-10 mV delta, I'll be a happy camper.

Unfortunately I don't actually have the required part values at hand so will have to go out and buy some on Monday.

Why Kenwood went to the effort of doing remote voltage sensing only to flub input voltage dependency (and thus ultimately load dependency), I don't know. I guess they were mainly focused on thermal drift and output noise, where the circuit is admittedly good. Ripple rejection also surpasses 80 dB by 50 Hz and goes well into the 90s, but the specifics may depend upon transistors there.

 

[AnalogSteph]

Update on that one - after retweaking the simulation (I think current variation is actually lower, like 175-250 mA, and 33 Vp with 15 ohms for each AC source seems more realistic) I have now settled on 68k-10µ-68k for the feedforward. 75k-10µ-68k would be a bit better still, but I bet I'm on the high side for my transformer source impedance estimate now. (I'm measuring about 10.3 ohms across both secondaries.) She'll be 'right, mate.

Parts have been obtained. (I'll have 56k, 62k and 68k resistors on hand.) Looks like I'll have some help with marginally better eyesight, too.

On a side note, this same exact regulator design with only minor variations in output voltage is shared by these other models:
KT-615 KT-815
KT-900 KT-1000
KT-1100

However, those with conventional pointer instruments (615, 815 and 1100) and as such largely constant current draw should be relatively unproblematic. The KT-900 with its even fancier LED S-Meter may be affected to a similar degree. It shouldn't take much to adapt the feedforward bodge.

(Before these, the circuitry actually was shockingly basic, indicating how quickly the prices of semiconductors were coming down in the 1970s. A KT-7500 used a zener + transistor regulator, a few years later relegated to the basic KT-50. In some later models, the problematic R115-116 feed is attached to a second regulated +28 V instead, in kind of a bootstrap fashion, thus eliminating the problem while improving regulator performance along the way.)

I may also address the topic of shielding of the absurdly long audio path in the KT-80. The output traces run literally halfway around the perimeter of the board. I had noticed some low-level buzz in the audio path, I won't be surprised if that's being picked up from the mains wiring. One more reason I want the square-wave-generating lighting power waster out of the way.

BTW, the KT-1100 strikes me as a case of the left hand not knowing what the right one is doing (something that seems to be common in the larger Kenwood tuners). Aside from the one (midly) reverse-biased electrolytic in the signal path that I found years ago, they're also switching the regulated supply for the FM LO quite crudely with a series pnp pass transistor (heaven knows what the tempco on Vce,sat may be), and I'm not sure whether the circuit for that is actually drawn correctly (they're using the base for another supply?!) - wouldn't be the first instance of such, the KT-80 audio wiring had some incorrect connections as well.

 

[Audiofire]

Impressive care that went to the tuner, I can sense the nostalgia

 

[Snoopy]

That's absolutely typical of the "high quality" Hypex amplifier PCBs. I have several here, all the same. That particular board is out of an NAD M-22 sent up from Melbourne.(Victoria).

Put anything under a microscope and it looks terrible. But yeah, those "SOTA" modules are pretty sh#t.

For real ? How is that stuff supposed to last decades ? And a NAD M22 as well ?
Ever look at a Yamaha pcb that close? I’m curious because I just ordered a AS2200

 

[AnalogSteph]

Impressive care that went to the tuner, I can sense the nostalgia

At least you can still work on this stuff without a microscope! I've been meaning to sort out the lighting situation since my days as an EE student when I got the unit, my first LTspice simulations would pretty much be old enough to drive now. I've always had a soft spot for better-than-stock mods on vintage hi-fi, but would usually help other folks with theirs. Back in the day I had a modest tuner collection going. I basically made the jump from shortwave to FM and then to music from there.

But impressive? Dragging out a VNA and hand-matching a set of IF filters to go in sockets, now that would be my definition of impressive. (A mod like this has actually been documented, see attachment - it is, however, in German. I don't remember where I got this back then, possibly the FMtuners group?)

The KT-80 was never my best tuner when it comes to handling high RF levels or separating stations under tough conditions (I had an Onkyo T-4650, Grundig T 7500, Kenwood KT-1100 and a very banged-up Grundig FineArts T-903 to compete), just a basic, good sounding (after taking care of the very dead 10µ/16 output coupling caps) compromise tuner that's mildly technically interesting on account of being the cheapest model with a pulse count detector, was put together with a certain level of quality, sports a nice weighted flywheel tuning knob and looks quite nice in silver. (Kenwood had no shortage of relatively unattractive black boxes later on. I've seen really decent models sell for shockingly little lately.) Also, the lack of an AM section keeps complexity to a relative minimum. With a simple wire antenna limiting RF levels, even the rather generous LED S-meter becomes decently discriminating, and the pre-cable frontend is very unlikely to struggle.

I've been considering getting a little '90s Onkyo with RDS as a modern complement to keep it company, after encouraging results with the micro sized T-405X (as discussed in my "kitchen-fi" thread). I would still like a KT-900, which is a really good-looking tuner in the dark in my book, but people seem to want almost as much money as for KT-1000s (or more) for these precious antiques.

EDIT: Man, the schematic and board layout for the KT-900 as per service docs are a mess. There seem to be multiple power connections in unswitched and switched B+ missing, and choke L2 next to the frontend is being shorted out. Either they did this on purpose or this thing was really thrown together. The real thing is decidedly lacking bodge wires, unless they happen to be on the bottom side.

 

[AnalogSteph]

KT-80 mods to lighting and regulator have now been completed. It's never as easy as you think it is (e.g. getting the power supply board in and out is a really fiddly affair as clearance is super-duper tight and the wires are a bit stubborn; you better get some Menthos for the hare's breath), but the good news is that all of them work just about exactly as simulation said they would.

1. Lighting

a) Part 1 - Power supply work
Old D24 was pretty toast, as suspected. Once removed, multimeter reading parallel to bulb went into the megohms. (Stubborn little fella, wouldn't readily come out.)
Old C90 actually still tested quite alright. (C = 1005 nF, ESR = 6.5 Ω, Vloss = 1.7%)
R111, as toasty as it ran, still measured 177 ohms. Not all heroes wear capes.

D24 was swapped for new small 8V2 (BZX23 series) installed a few mm above board height. (According to sim, Pd = 0.06 W with LED bulb replacement installed or 0.1 W without. It's a nominal 0.6 W part.)
Added rectifier diode Dnew (1N4002) was installed below, its parallel capacitor Cnew above the board.
R111's remaining legs were bent until they could be soldered to my 22.5 mm 3.3µ/100V laying down.

Result: 8.138 Vdc and a steady green dial pointer LED.

b) Part 2 - Power indicator LED mod
The dead bulb could be pushed out of the silicone holder thingy and cut off. Holder thingy ultimately was not reused and the cable now comes in from the left and above with the LED at the end pointing towards the responsible lightpipe area. Warm-white LED turned out to be one sans integrated resistor after all, so I found two 2k2s in my stock which were used in parallel. Heatshrink of various diameters was used. The whole shebang was fixated with hot glue.

I think the warm white LED I bought is a type with the following specs:
Mfr = OptoSupply
Vf = 2.9-3.6 V
If = 20 mA
dia = 5 mm
light output = 4200-5800 mcd
angle = 60°
lens = transparent
chroma coords = x: 0.45; y: 0.41

Anyway, it's plenty bright enough and a fairly nice warm white. You could still go with half the current instead, I think. I may have gotten lucky in that it's a relatively wide-angle affair. Diffusing the housing may still be worth a shot, but it's not terrible as-is.

EDIT: With both LEDs working, LED supply voltage is 7.975 V +/- 5 mV. A voltage swing of about 38.2 to 39.8 Vac (about 4%) is being reduced to a 10 mV difference (about 0.125%). That's better than expected and, in fact, better than the main regulator's performance in stock form! There's about +10...15 mV worth of initial warmup drift, then that's pretty much it. Precision lighting.
The dial pointer LED by itself drops 1.95 V. It must not be very stressed at 6 mA.
AC wise, the multimeter states 0.2 Vrms. I still notice a hint of flicker on the dial pointer LED when contrasting against the S-meter LEDs. The white LED with its slow phospor is perfectly happy as-is.

Off but plugged in, LED supply settles at 1.63 V through leakage from spark killer cap parallel to power switch (with absolutely no light output whatsoever that I can see, currents must be absolutely minuscule). Once unplugged, it slowly decays further.

2. Regulator

Values used: 68k 1% - 10µ/63V - 68k 1% (plus some thin solid-core wire for the Cff-Rff2 connection and a dab of hot glue to hold it). Everything installed below PCB. Rff1 and Cff at R119/120 and the ground-bearing jumper wire next to them, Rff2 at D16/17 junction.
New Vout = 14.23-14.24 V. Slight negative tempco by the looks of it. (Not a real surprise since effective Vref includes Vbe(Q11) with its usual -2 mV/K, times about 2.11.)
Vout,maxload - Vout,minload = +1 to +2 mV. Maybe we could have gone for 75k for Rff1 after all, but either way, it's darn close to load invariant. A massive improvement over -92 mV for sure! And all that with 3 passives. Suffice it to say that I am not going to pursue the current source avenue any further. Simple and effective is what it's all about.

The toastiest part on the power supply board now is the regulator series transistor heatsink, as it should be. I have seen 53 °C on the thermal camera, which is not super great but acceptable. I have seen unregulated voltage reach a hair over 24 V unloaded and about 22 V at max load (though that was before installing the LED), so it has to drop close to 10 V now. I suppose you could source another 12-0-12 or 15-0-15 transformer if you insisted, but the question is how much gilding the lily really needs. I also suspect that my 3.3µ dropper may appreciate some series resistance when going with a toroidal.

Power consumption at 230 V (will vary slightly with voltage):
No antenna, no S-meter LEDs: 5.8 W
Full S-Meter, stereo, AFC: 8.4 W
Power factor: 0.87-0.90

Quite an improvement over the previous 10-12 watts. With another transformer you could probably more than halve the stock power consumption.

Frequency drift as a result of lower voltage has been minimal at best. (EDIT: Possibly because the oscillator appears to be of the emitter follower Colpitts variety, with external capacitances dominating internal parasitic ones. This seems to have been a common choice in analog tuners.) Readout still is about 0.1 MHz high near the top end. Thoroughly cleaning the green oxidation off the varicap wipers with an old toothbrush and Video 90 spray (no-residue VCR head cleaner and pretty much the same as tuner spray to my knowledge) seems to have brought the bottom end back into spec pretty much exactly. It definitely eliminated the scratchy tuning in any case.

The toastiest part on the main PCB seems to be the RF preamp MOSFET. The SC114 hybrid isn't even that hot but would also seem to have decent thermal connection to the varicap.

Thanks go out to local electronics wizard Peter for all of his help. The whole process still took close to 4 hours in total, though our primary focus wasn't on hurrying either. As usual, doing it again would go more quickly. Likewise, a routined electronics tech could no doubt speed up the process a fair bit as well.

EDIT: The glory shot - well, actually it wasn't particularly glorious, natural light hasn't been super great the last couple of days. I've made it as pretty as I could under the circumstances.

Below: Mods on the power supply board, top side.


Update 2: I found a spot with a bit of new white residue near C82 on the power supply board 2 weeks later, and there also is a slight urine-like smell which seems most prominent in the filter cap area. Methinks the poor old filter cap may have gone slightly incontinent, even if I wouldn't say the smell is very fishy. Subjecting a 45-year-old 25 V type to a hair of 24 V did strike me as a bit risky, I will admit. I have a 1000µ/63 on hand that would be the exact size of the original (16x25 with 7.5 mm lead spacing), otherwise I could get 3300µ/50 and 4700µ/35 types in 18.5x35.5 locally that I reckon should fit as well... the slightly wonky C79 may need some persuasion, mind you.

Side note, there is no shortage of bloopers among Kenwood tuners and their service docs. The KT-900 has to be one of the worst and screams "rush job" (nonexistent power connections, L2 bridged over, "wide" switching diodes running at like 3x the current they should - I have one of them bug collections on the way now, let's see how bad it really is). But even the KT-815 alignment instructions for the "noise amp" refer you to a phantom JFET Q3 (still listed on schematic but nowhere in sight) and phantom diode D8 (nowhere in sight either). And I guess only the designers themselves will know what the business with Q16 at the regulator is all about in the KT-1100. All of this makes the KT-80's misdrawn output connections looks downright negligible.

 

[AnalogSteph]

Looks like one of my AM/FM/shortwave portables has given up the ghost while in the box - a Roadstar TRA-2350P (Redsun RP2000 rebadge) that I bought in 2006. Completely dead. I found +3.3 V on the goldcap, so it's not that. Likewise, the classic electrolytic in the reset circuit seems to be charging up fine. So my bets are on a missing clock now. I noticed that the µP has two pairs of clock pins, XTIN and XTOUT with a 32.768 kHz clock crystal (so no doubt for timekeeping) and another named XIN and XOUT merely connected by a 3.3 kOhm resistor. What is that, a random frequency oscillator? Not exactly looking forward to removing the shielding. At least there's service docs for the thing, so there's that. If I do get it running again, I guess I'd also address the puffy filter cap and maybe even the very hummy transformer (finest 2000s era China quality right there), alongside the broken handle. This model wasn't very expensive at all for a larger portable, and it shows.

On another note, I was saddened to find out that Murata's IF filters have been dead and gone for almost 2 years now (and I mean all of them - maybe because PZT is a lead compound?). That figures, now that tools to characterize them have become easily affordable for the hobbyist (I was looking at a 1.5 GHz nanoVNA for <100€), the darn things go extinct. Reportedly the Chinese makes remaining are no match in terms group delay and insertion loss. The high-performance FM tuner as a mass-market commercial product and an art form has arguably been dead since the 2000s, but now it's truly dead and buried.
Well, unless you've got the deep pockets for an Accuphase T-1200, of course (6 grand as usual). They can afford a custom dual conversion DSP job. (9.216 MHz IF1, 3.072 MHz IF2, digitized at 24.576 MHz / 16 bit, processed by DSP1, then decimated and processed by DSP2 before sent to an AK4490 at 24/48. Up to 92 dB SNR is as good as anything ever.) The rest of the world can consider themselves lucky if they get something 1-chip from the automotive world like a TEF668X (which seemingly are really great little chips and kind of have to be to be viable for a car radio, but still) in a portable kit-like affair. As far as I'm concerned, a tuner manufacturer could be dressing up one of those with a nice case and weighted flywheel tuning knob + optical encoder and it would be a viable proposition. Not exactly a new idea, the Onkyo T-4030 (2012-2021) used a 1-chip Frontier Silicon Kino 4 (FS1445-A) solution already, and in the legendary Sony XDR-F1HD (2008) it was something NXP.

 

[AnalogSteph]

I finally managed to snag one of them elusive Kenwood KT-900s; lowballing a seller (apparently an antiques dealer who'd taken the auction photos all the way back in 2020) did the trick. Save for a strange dust accumulation behind the front glass (how'd that even get in there, the unit is largely airtight?) and a cover that could arguably use a respray as they often do, it's in decent condition, even the AM antenna and a user manual were included. The cereal number (203xxxxx) may indicate a 1981 unit, which seems right as I don't think these were very long-lived. (My KT-80 S/N 1xxxxxxx still has its late 1980 warranty sticker on the bottom.)

Initial assessment:
No stereo in sight despite seemingly strong rx. (Must have been that way for a while, too, as I got the unit switched to mono.) Rec Cal doesn't say a peep. Frequency display jumping all over the place when tuning. Frequency counter often losing signal towards the bottom end and defaulting to 89.3 then. Dial about 0.3 MHz low.
(Later: ) AM tunes all the way to 1871 kHz and is a tad deaf up top.

Measures taken:
Tightened a number of loose PCB screws.
Attacked varicap wiper contacts with contact cleaner and toothbrush. No more jumpiness.
Found electrolytic C93 (-18 V) was making contact with hot resistor R181 (~100°C). Which sadist assembled this unit?
Found stereo decoder VCO adjustment way out. Tweaked the pot until stereo returned, followed by fine tuning to ~75.5 kHz free-running with the little-used frequency counter function of my multimeter. (The HA12016 datasheet suggests a maximum of separation in the 75.1-75.6 kHz ballpark.) Must have been 72 kHz or even lower originally. Tip: Do this adjustment with Rec Cal on, the oscillator is a lot less jumpy with no noise in the background. The pot remains twitchy.
Rec Cal again needed a major pot tweak to come back alive. Set to about 0.20 V as we're in 40 kHz deviation terrain here, and that seems good. (Instruction were 6 dB below the level for nominal deviation.)
The LO capacitor adjustment always is a "tongue at the right angle" kind of affair that takes several tries to get right. I eventually settled for a +0.1 MHz dial offset towards the top as the bottom end was about as much low in return and didn't feel like messing with a stubborn air coil.
AM LO adjustment is still twitchy. You could go all the way up to 2.3 MHz if you insisted. Settled on 514-1655 kHz total coverage. The area around 900-1000 kHz is 2-3 kHz out but the rest is pretty much bang-on. (AM actually sounds like it has heard of selectivity for a change, though it's not exactly wide bandwidth.)

This is the filter complement us Eurotypes got:

I'm blue, da ba dee da ba di...

Wide: SFE10.7ML x2 (280 kHz GDT, same family as MM, so not the real fancy MX family jobs but I don't know whether they even existed yet)
Narrow: those plus two SFE10.7MJ 150 kHz non-GDT types (note how those are of different height).
All at 10.67 MHz nominal.

Surprisingly, wide gain adjustment seems to be spot-on as found. (EDIT: Scratch that - but it only needed a minor tweak.) I have no immediate complaints about the discriminator either, and the 2nd LO seems to be bang-on at 8.71 MHz, too (=40/49*10.67 MHz). I would like to check stereo seperation anyway.

The disparity between voltage across R7 in wide vs. narrow is not as great as I had feared, being about 4.8 vs. 3.0 V. R9 is still too low IMO. EDIT: Oh yeah. R14 drops 8.34 V in Wide vs 3.14 in Narrow. There's 12 V after R9, coming from a 12.8ish V supply.

Frontend supply at L2 (the infamous bridged-over inductor) seemed lower than the usual 13.5ish V. (Which have a maximum load variance of about 80 mV, on account of all the lightbulbs still working.) I'll have to evaluate how it even gets there to begin with, which is not at all obvious from service docs (which are blooper city - I stumbled across an old post of mine from 2006 where I had found yet another one, the left and right channels get reversed at some point; I hope the actual unit is OK...).
EDIT: Yes, channels seem to be correct IRL. Also, the current of frontend and IF strip is run through diode D5 (not in schematic but below J20 in layout), which explains the drop to 12.8ish V.

Isn't she a sexy beast?

Static pictures don't really do this one justice; the color-changing S-meter bars, tuning indicators of varying intensity and backlit inverse dial plus knob touch detection automatically defeating the Lock function (i.e. like the KT-1100 has) make for quite the lightshow when tuning around. Should probably take some video. And yeah, the S-meter is very animated - more things are than aren't a 7. I'd estimate that a 2.5/5 on the KT-80 might be a 4/7 and a 3-and-a-bit/5 a 6/7 here, so effectively we haven't gained much. (I have never been very happy with the S-meters in my Kenwood tuners, unlike say a Grundig T7500.) Anyway, using this unit is quite the event, in ways that we might rather associate with the enthusiast automotive sector these days. The tuning knob sports an even bigger weight than the KT-80's, too. This is how you design a product. Only the dial pointer light wiring dragging across the dial string is a bit disconcerting-sounding. And I wish the button positions aligned with the respective indicator LEDs.

The unit is surprisingly large - while slim, the case is very deep, deeper than either the KT-80 or KT-1100, plus AM antenna at the rear.

At between 13.5 and 19.5 W from the wall (AM or rec cal vs. FM with everything lit), things are getting quite warm in the long run. A lot of that heat is from the dial area or from series resistors associated with bulbage.

The single one with 220 ohm series resistor R181 presumably is another 8V/50 mA similar to what the KT-80 had for its power button. (Note, most of its light is actually wasted. That spot would have been perfect for an LED.) It drops 11.4 V IRL, suggesting 52 mA, we can blame the newfangled 230 V mains for that. The bulb has a slight greenish-blue tint, not sure how much that has faded over the years. (I don't think the unit is super high hours but still, it will have seen some action.)
The others are in series with R180, which drops 2.68 V, suggesting 149 mA, so these are probably 8V/150mA bulbs.
In sum, same types as used by the KT-1000, no surprise there. And about 3.8 W secondary-side if I'm not mistaken. Could be 5+ W primary-side then.

Finally, a protip: If you get a random old piece of wire from the stash for an antenna, at least give it some contact cleaner first, 'K? Otherwise you might be greeted with a surprisingly amount of IMD from the local flamethrowers.
EDIT: Nope, looks like the unit's strong-signal handling really is a bit naff, definitely worse than the KT-80's. That's odd, after all the circuitry is virtually identical.

BTW, the 300 ohm screw terminal on the right seems to give me better signal levels with a random wire than the other one, easily 1 bar. Peculiar.

Update - Performance notes.
Sound is nice and clean and all, but you know what? She doesn't have the bass slam of my KT-80. Now granted, the combination of her 2.2µ output coupling caps (in unknown shape and with zero DC bias for nearly 45 years) with my headphone amp's 10 kOhm input impedance is not exactly ideal, adding to the impact of 3.3µ and 6k6. And unlike in the KT-80, you can't just bridge them either.
EDIT: Well, I'll be... Here's both compared with a mono signal using my EVO 4, input impedance spec "10 kOhm balanced". Not even .5 dB at 40 Hz?

LO cleanliness seems roughly on par with the KT-80, maybe a hair more hum. She's drifting quite merrily though... same order of magnitude as the KT-1100, but more steady than erratic.
Aligning the LO brought AM RF tracking back in line, but levels still drop towards the top end of the band. I am starting to suspect that this might be a hardware limitation.
Dimensions would suggest having the KT-900 on the bottom of the stack, but the case is pretty flimsy and you shouldn't really stack much of anything on it.

Among reassembling the unit, the VCO alignment seems to have drifted yet again, grr. (EDIT: Nope, false alarm, stereo just needs 4 bars of signal minimum on this model. But manhandling the unit again has apparently shifted my FM 1st LO adjustment by 100 kHz.) I think those pots are a bit crap, the KT-80 had nicer enclosed ones. I noticed that this one turns quite easily. There are also some spots with dragging tuning. EDIT: Yeah, that's from the big reel on the side of the variable capacitor, which fits with microns to spare. You have to assemble the case very carefully and shift the lid just right to avoid dragging. Pretty lousy construction - all thin sheet metal (not "cut yourself when looking at it wrong" thin yet, but still), poor interlocking, and then those kinds of tolerances. Makes the KT-80 look like the proverbial brick outhouse.

So - preliminary action plan:
Address the coupling capacitor situation. (Fake bipolars?) - EDIT: Gave the old output caps some voltage for several minutes via my multimeter's diode test, and at least my imagination would want to claim that it has made a difference.
Maybe some more recapping in the power supply section.
Regulator mod à la KT-80. Have a look at thermal stability. I noticed that the voltage seemed to have a more pronounced positive tempco. D5 would give the oscillator supply even more. I'll need to consider extreme measures such as including diodes in the reference divider to establish a basic negative tempco.
Replace VCO trimpot and maybe Rec Cal trimpot as well.
Clean the dust or condensation in the front, if accessible. (Getting the front apart to get to the lightbulbs and everything back together is supposed to be a bear, according to Hiro-Kun who's had no less than 8 Trio KT-900s on the bench plus several of the similar KT-990s [which seem to have been a Japan-only affair from 1982].)
An LED mod would be nice in the future. The unit makes a nice hand-warmer after running for a while.
HOW DID I NOT NOTICE THAT THIS GUY HAS NO MAINS FUSE?! (Well, turns out that this is actually the norm for Kenwood tuners, and the European version of the KT-80 getting one seems to have been the exception to the rule. Yes, the transformer has a thermal fuse, but c'mon.)

 

[Sal1950]

Isn't she a sexy beast?

A purdy ole girl, hope she shows you much love.

 

[Audiofire]

Nude photo thread!

 

[Mschn]

Musical Fidelity A-X2 to get new electrolyths some rainy day:

/Martin

 

[Palmer]

Revox B77 that I managed to snatch for cheap, hard life - left motor has a bad bearing, and the heads are quite used. Got most parts, and am working to bring it back up.

Found a picture of a very modified A77 on an Auction site, that was unfortunately already sold. If anybody knows what that is, I'd like to know - never seen one of those.

 

[AnalogSteph]

More Kenwood tuner woes:

Problem #1.

I noticed that my KT-1100 has a non-functional AM section (with mere hints of the LO briefly starting to operate), and the FM oscillator remains running in AM mode (although the frequency counter duly switches modes). That forced me to do some proper tracing of AM power, whereupon I found that my previous theory was incorrect. Q16 is merely being used to turn off FM frontend power when the AM supply comes up, the AM section's main power connection comes in via D15 and eventually goes back to the AM/FM switch. I reckon there's a bad solder joint somewhere. Great, now I have to drag out the tuner that sits at the bottom of the stack. That figures.
Update - 13.5 V before D15, 1.7 V after it (measured at the diode legs), and neither D15 nor anything in the vicinity is getting hot. That's pretty conclusive, it's a bad freakin' diode. Other than a slightly higher forward voltage drop of 0.68 V instead of the 0.61 V of its neighbors there's nothing obviously unusual about it. What an unusual fault, but hey, I'll take it.
Update #2 - she's alive! Bought some parts including several 1N4936 diodes today. Replacing a single diode can be quite a chore when the legs are bent over really good and you have to work around a stupid coax cable to boot, not to mention I never see all that well on a good day, but I got it done. (The original part measures normal again now, but I wouldn't trust it.) Really needed realignment, and not just in the antenna circuit that I would have expected (looks like the original antenna is different), even the LO was quite a bit off. It sounds much better and is more sensitive and generally louder than the more mid-centric KT-900, however I struggle to ever get rid of 9 kHz hets even in Narrow. Looks like a barn door plus audio filtering does not good selectivity make. (The IF filter is an SFP450F, a 12 kHz job.) If memory serves that had been a complaint of mine even when I got the unit, so the AM section must have worked then.
Now how does one adjust the play in the tuning knob bearing...?

You reckon one could use a MOSFET of suitably low Rds,on to replace Q16 as a switching transistor? A p-channel, obviously. Heaven knows what the tempco of BJT Vce,sat is (apparently positive), even if specified voltage drop is supposed to be only about 0.1 V.

Problem #2.

My advice about using contact cleaner on the antenna wire may have been premature... looks like the KT-900 really does have pretty naff strong-signal handling. (Maybe that's why it got a so-so review score at the time.) I can get it to produce fairly strong IM3 with just two pieces of wire on the 300 ohm terminals. Granted, RF levels seem to be fairly high in my local urban environment (my poor old Panasonic RF-1410 kitchen radio was always struggling), but still, it's not like I'm subjecting it to cable levels (which I can't test as analog FM on cable was turned off a while back). That was a bit of a head-scratcher, after all the circuitry seems virtually identical to the KT-80 which is substantially less easy to impress. The mechanical varicap obviously has the AM plates populated that were absent in the KT-80's, but otherwise I reckon it's the same basic part.

Strike 1 - B+. Where the KT-80 supplies its frontend with 14.9ish V stock and 14.3ish V after my mod, the KT-900's only gets 12.8ish V to begin with. That can't be helping. Not entirely sure why they put the frontend and IF strip behind a diode (if it's needed to avoid reverse discharging, I'll be considering putting in a Schottky instead), and why regulated voltage is only 13.5 V to begin with. (I suppose it's closer to 14 V than 14.9 was, so some progress has been made.)

Strike 2 - devilish details.
KT-80 frontend:

KT-900 frontend:

A lot of things are the same - semiconductors (although an LO buffer is added for frequency counter operation), FM gangs, Balou the balun (L19-0026-05), resistor values. Coupling caps are a bit different, but who needs a 100p in front of the RF preamp in this frequency range anyway. LC fixed capacitors are bigger in the RF circuits and smaller in the LO, but since actual variable capacitance has to be pretty much the same and so is the range to be tuned, inductances can't be much different and so Q shouldn't be either.

One thing struck me as peculiar: The KT-80 gets a 1.5p coupling capacitor for the LO, while it's only 0.5p in the KT-900. That has to be negatively affecting LO levels at the mixer, which in turn would be worsening its strong-signal handling capabilities. My hunch is that the LO buffer Q2 via its 1.5p may have been posed enough of a load that it kept the LO from operating reliably. Also, why does R7 have to be as low as 1.5k? Something smells of "let's bodge it until it works" here.

Great, now how does one fix that? Do you reckon one could take some hints from the KT-50, a basic little tuner with a frequency counter that used to sit under the KT-80 in Kenwood's lineup? It uses the same SC114 hybrid but only runs on 8 V, which I guess is what necessitated a bigger value for C7:

C8 = 1p certainly sounds more promising. Also note R3 = 100k while using the same type for Q2.

Update: If it's any consolation, the little Onkyo T-905X seems to struggle more than either of the two Kenwoods, presenting a clearly audible IM3 mixing product when things have quietened down on the others. Another basic 4-gang frontend (this time varactor tuned, so likely lower Q) with a bipolar mixer and no RF AGC.

 

[Mschn]

Musical Fidelity A-X2 to get new electrolyths some rainy day:
View attachment 452330
/Martin

Rainy day came and went. All electrolyths have been replaced. The re-assembled amplifier seems to work OK so no harm done.

What else can I do to this old amplifier? I don't have schematics but I do have a data sheet for the power transistors (Sanken SAP15P/SAP15N). I do have electronics lab experience but only microwave (1 GHz and up) so not very relevant. The only measurement instrument available to me these days is an old Fluke 8060A True RMS Multimeter.

Any ideas on how to assess the fidelity of this amplifier? The only thing I can compare with at home is a not quite so old yet old Denon AVR-2105.

TIA

/Martin