Mitsubishi DA-P20 Preamplifier Measurements

[MAB]

I dunno... the dominant harmonics seem to be H3, H5 and the fundamental, with the even orders you'd expect from a bridge rectifier (i.e. 120 Hz harmonics) trailing behind. The fundamental may be down to leakage current, but H3 and H5? And yes, I checked, only bridge rectifiers in this one. If the measurement setup passes the null / sniff test (H3 and H5 famously being those not cancelled out by a balanced input), it may be time to look at the mains wiring inside the device and which high-impedance nodes might benefit from some extra shielding.

Thanks Steph. You always give positive feedback! Really interesting point.

Looks like I am under tons of pressure to recap the thing first, fix it later. I see no other evidence of capacitor issues despite the age.

BTW, the heck is this MC prepre?
View attachment 449867
Surely this would not be an inverting amplifier with a 10 ohm input impedance? Well, I'll be... it's right there in the specs, 10 ohms. I guess with MC carts being mostly resistive, you can pull it off, but wouldn't that be a bit less than ideal from a noise perspective?

I have a Marcof PPA-1 MC head-amp with 10 Ohm input impedance. I need to dig it out.

I see someone discovered the "power transistors for low rbb'" trick a long time ago... but of course there's no rbb' data for either 2SD467 or 2SB561 out there, so without having some on hand nobody knows how suited they really are.

Thanks. Very interesting.

 

[Bruce Morgen]

Early Life Fails in electrolytic capacitors is quite high, much higher than end of life fails.

Not in my experience and certainly not for any of the known-reliable brands -- but of course YMMV.

 

[Bruce Morgen]

Just a lighter moment for techs and tinkerers in their "golden years."

 

[Sokel]

I remember someone searching to change the already changed back in the day 2SA1360/2SC3423 which are of course also obsolete for some time now.
Good luck with those, don't know with what they swapped them with.

 

[MAB]

Just a lighter moment for techs and tinkerers in their "golden years."

Very nice!

 

[Dougey_Jones]

I didn't watch the entire hour.

A couple points:

1) No way I would blanket recap the unit as he did. The practice of blanket replacement of components actually introduces unreliability for either zero or (at best) random improvement. He states some facts about capacitors that I actually agree with, but then leads his viewers to believe that a capacitor that is 'obsolete' is somehow inferior and in need of replacing which I disagree with. And if you do recap, a before and after set of measurements is needed so you know what you actually fixed, and what you broke, and what was already broken. I 100% disagree with recapping as a rule. There seems to be an active industry in selling recapping services since it generates revenue for the simple task of part-swapping.

2) Regarding already broken, there was a pre-existing problem in the unit he serviced, the faulty relay. He part-swapped the capacitors in a unit with a bad relay. This is the wrong order of approach.

He should have thoroughly checked out the unit first, identified the problem(s) and fixed it, and not tried to fix stuff that isn't broken.

45 year old caps are a no for me, dawg, but you do you.

 

[sc00ter]

I have an old Mitsubishi DA-P20 preamplifier. It was part of a modular lineup of separate components.
View attachment 449314

The preamp could be bolted onto a couple models of matching amplifier:
View attachment 449316

The resulting Franken-Receiver is slightly unwieldy.
View attachment 449315

Fortunately Mitsubishi had a pair of VU meters you could bolt onto the amps.
View attachment 449318

The matching DA-F20 FM tuner is remarkable and I still use to this day:
View attachment 449319

My preamp has had a bumpy life though.
View attachment 449321
I've loaned it out a few times. It was used for a business for a couple years. It's never been serviced. It has no obvious faults, and even the stepped volume attenuator, tone controls and switches have only the faintest noise. It has dual tape loops, MM and MC cartridge inputs, and boasts a 'dual mono' design with separate left and right input attenuators, plus separate tone controls for each channel. It boasts some killer specs.
View attachment 449323
18 Volt output, 10 Hz - 100 kHz frequency response, 0.002% THD for line-level devices, 290 mV phono overload for MM. I'm going to test as much of this as I can, compare to Mitsubishi's published spec. I'll also provide a similar dashboard of test conditions that Amir provides. I measured the unit with a QuantAsylum QA403. I measured, cleaned and adjusted Left and Right rail voltages to spec. There is a slight channel imbalance, more on that below. I was able to compensate for the imbalance with the handy input attenuators.

Feeding 2.5 V 1 kHz, gain set to provide 2 V output similar to Amir's dashboard:
View attachment 449328
This is great performance. The above is with the tone controls defeated.

With the tone control switch enabled and response in flat position there is a small 2dB degradation in performance:
View attachment 449330

Sweeping input voltage at unity gain results in the following output THD and THD+N:
View attachment 449332

I think that is outstanding. I cranked it all the way up to see if it can really deliver 18V :
View attachment 449334
It does put out 18V, even into 1 kOhm! You need a hot source, but it will do it if fed enough input voltage! I think you can spot weld with this.

Mitsubishi claim wide bandwidth:
View attachment 449336
I measure -0.2 dB at 10Hz, -1dB at 80 kHz; so slightly below spec but still great. Green trace is with tone controls defeated, the yellow is with the tone on and the controls flat:
View attachment 449338

I don't know if this is overpromising or due to age. The FR degrades when tone is on, but still impressive.

The tone controls have the following response for the +-2 dB, 6 dB, and 10 dB settings on bass and treble:
View attachment 449346

View attachment 449345

The tone filters roughly conform to the spec, have 2 dB increments.

The subsonic filter is different than spec:
View attachment 449348

Mitsubishi claims -6dB at 18Hz, but slowly rolls over an hits -6dB just below 10Hz.

Here is the performance of the volume control:
View attachment 449355

Maximum gain on the left of the chart is 16 dB. The largest left / right imbalance is at moderate listening volumes, in the range commonly used.

The stepped attenuator staircases in this region:
View attachment 449354

I cleaned the volume control, it didn't help with the channel mismatch or the staircasing. I measured the resistance of the ladder at each volume position, it tracks exactly the L and R gain. The L/R imbalance is due to the attenuator. I don't know if the resistive properties have changed over time, or if this is the same performance as new. It's still good performance, and the small amount of scratchiness that had developed is now gone. I am able to level the channels with the input attenuators for the measurements, but that isn't practical in real world since the imbalance changes with each attenuator position. It's not bad, but the nonlinearities in the volume control may be the most audible artifacts of this preamp since the rest of the line level performance is so good.

I'll test the MM and MC sections next. For instance, I want to see if it meets the 290mV overload spec:
View attachment 449356

Nice! I had the DA-F20 FM Tuner (great tuner, and really good-looking), gave it to my nephew’s 15-year-old daughter along with a turntable, integrated amp, and some other stereo gear when she got into vinyl a year or two ago. The Mitsubishi DA-R10 was my first piece of decent mid-fi equipment years and years ago, moving from Panasonic to Technics and finally Mitsubishi.

 

[EJ3]

Blanket recap is a bad practice because of the bathtub curve.
Early Life Fails in electrolytic capacitors is quite high, much higher than end of life fails.
If it was lower, and the rework is done flawlessly, then replacement makes sense if you have the time and money.

If it was only 10 years old, I'd agree. But when you are starting with a product that is 35 years old, I'll disagree with you.
Even the article says the following:
This article relies largely or entirely on a single source.
Not all products exhibit a bathtub curve failure rate.

 

[MAB]

If it was only 10 years old, I'd agree. But when you are starting with a product that is 35 years old, I'll disagree with you.
Even the article says the following:
This article relies largely or entirely on a single source.
Not all products exhibit a bathtub curve failure rate.

Actually the bathtub curve is the accepted model for a capacitor. I work on reliability testing of capacitors in semiconductor systems, across a range of temperature and voltage stresses, bathtub curve applies for sure. And is used widely to describe capacitor life expectancies in the aerospace, medical, and transportation industries, to name a few. Weibull described and modeled the phenomena in 1939, his distribution has given good service in modeling capacitors, bias temp inversion of transistors, incandescent light bulb failures, and many other systems since. And all of the major capacitor manufacturers use it to guide the customer through the life expectancy calculations of the product in their application notes. In fact, of the capacitor types, electrolytic caps are about as close to bathtub as any. This was way outside of the scope of this thread which is a review of a preamp, but since it was asked, here are just a couple of manufacturer's application notes with detailed discussions of the bathtub curve:

https://www.cde.com/resources/technical-papers/AEappGuide.pdf

https://www.nichicon.co.jp/english/products/pdf/aluminum.pdf

This thread is a measurement vintage gear measurement, not a blanket replacement of capacitor discussion, or a reliability discussion. I regret it keeps getting drug into the weeds. The capacitors aren't this preamplifier's problem anyway. It does have a couple of issues. One is the performance of the volume control that I mentioned in the first post.

In fact the preamp used first-class caps for the time, operates them well below the voltage rating, and most importantly operates the caps at just above ambient temperature. Which if we follow the rel estimates provided in the above links, you will see the lifetime derate due to the low operating temp gives wear-out estimates of decades. And if the are degrading, it isn't some catastrophic issue like many other component.

Aluminum electrolytic infant mortality rates are actually quite high, and represent a greater risk to rel than an aged capacitor so long as it is operated below it's max voltage, and especially max temperature. Critical military, transportation, and medical applications specify multiple layers of acceptance testing to prevent components like new capacitors from introducing infant mortality fails into freshly repaired systems. This is just a piece of audio equipment so the stakes are really quite low, and if a cap goes bad in the next decade it isn't the end of the world, and the caps are not degrading the sound or operation in any way shape or form.

 

[Bruce Morgen]

FWIW, in my eight years managing a factory-authorized service shop for 50+ brands back in the 1970s, the number of warranty repairs -- obviously including any "infant mortality" period -- traceable to an electrolytic capacitor failure could be counted on the fingers of Django Reinhardt's left hand.

 

[MAB]

FWIW, in my eight years managing a factory-authorized service shop for 50+ brands back in the 1970s, the number of warranty repairs -- obviously including any "infant mortality" period -- traceable to an electrolytic capacitor failure could be counted on the fingers of Django Reinhardt's left hand.

Perhaps blanket capacitor replacement requires it's own thread.
I don't make money off of myself for recapping this.
Nor do I make money off of any restoration or repair.

I would also like to stop wasting time on this thread, which isn't about a blanket cap replacement.

 

[SegaCD]

Actually the bathtub curve is the accepted model for a capacitor.

The bathtub curve widely reflects most forms of manufacturing that create devices that perform under some kind of stress, not just electronics. Failure rate at the outset of a device's useful life almost always far exceeds that in steady-state use. I agree its a great reflection of how the initial stresses of use are usually different than the stresses in manufacturing.

The thing is that most manufacturers have mechanisms in place to reduce the initial issues along the bathtub curve. The CDE document you shared explicitly states that they pre-burn-in their caps to reduce the customer burden of the bathtub curve, just like a steel manufacturer may radiographically test the structural hardware they produce. As someone who works on hardware and is obviously concerned with a functioning product, you also know that failure rates of your final products are a sum of all the failure rates of all the components on your board. This means that your components should be pre-tested or pre-burned by the manufacturer to reduce their failure rates to a minimum, particularly the "Infant Mortal Failure Rate." And this is key takeaway here; the "Infant Mortal Failure Rate" of the bathtub curve is greatly reduced by manufacturer steps in their QA process. It is not eliminated, but it is highly reduced. In addition, such failures which occur on the field are usually characterized and bad-batches are usually caught & communicated if necessary. I would also argue that initial mortality of passives are usually (but not always) catastrophic. This doesn't always mean the capacitor will explode, but usually it's a steady-state failure making it fairly easy to weed out.

But the failure in your argument is the lack of acceptance of time & usage cycles as a factor. If you're working with MIL-spec hardware, this may not be a concern as components will usually be left powered on, inspected/calibrated against limits as part of regular maintenance, and carefully monitored if idle for extended periods (and I assume that most hardware decades old will either be in constant maintenance or retired), but when electrolytic capacitors exist for years and years with a nebulous quantity of use and storage, aging calculations become a lot harder to quantify. In both of the documents you provided, the calculations do not accept rated shelf life as a parameter. All the aging calculations are "in use." Even so, look at what the Nichicon document explicitly notes as a maximum life...

...15 years! That's right; they are saying "Yeah, you really shouldn't exceed 15 years of use on ANY line of our capacitors." So when you have a ceiling of steady state "on time" of 15 years, a rated shelf life of 5-10 years, and you use your amplifier (or certain circuits of your amplifier) periodically under varying load sometimes for hours at a time & sometimes not for years, how old does your capacitors need to be where the risk of wear/age-related issues in the bathtub curve begins to exceed acceptability of the risk of initial morality? Can such a thing be characterized or calculated? I haven't seen a reasonable capacitor aging algorithm/study for medium to low random use over many years. Could it be greater or less than 15 years? From what, I see I can not see it exceeding 15 years.

And then, on top of that, while initial mortality is largely characterized, I must reiterate that its very hard to characterize age-related issues from irregular use. Your Nichicon document's "analysis of failure mode" diagram has almost every failure mode as possible under "time-related deterioration" and it's true. I have seen incredibly bizarre behaviors from capacitors that are 30 years old and have a fine measured ESR and capacitance as I previously mentioned. It is really not worth "guessing" if a capacitor's life is up; it's much better to assume. If you're worried about infant mortal failures, then go slow and replace board by board and backtrack if there are issues. Of course, if you go to a money-is-the-bottom-line tech, that sort of process almost certainly won't be possible... If this isn't implied by my thoughts so far, let me explicitly say that if you own 30+ year old electronic gear, you need to know how to work on it yourself or have a very close tech friend. Otherwise, it'll just be a burden or you'll be left telling fooling yourself that the hardware issues you encounter is just how it was "back in the day." I used to believe the latter until I got fed up opening my hardware every 6 months to a year for maintenance. Now I go years without major issues (environmental issues aside).

While this is a pre-amp review, it is a vintage pre-amp review. Its important for folks to know what was and wasn't done to the hardware and what that can potentially mean to the measurements. I've submitted a few pieces of vintage hardware to Amir (like the Yamaha A-1), and I've seen the effects of leaving working but aged capacitors in place (like an elevated noise floor). Vintage hardware like this Mitsubishi is incredible, but it can be even better with proper restoration. I hope that this conversation can help the review, not detract from it.

 

[EJ3]

Actually the bathtub curve is the accepted model for a capacitor. I work on reliability testing of capacitors in semiconductor systems, across a range of temperature and voltage stresses, bathtub curve applies for sure. And is used widely to describe capacitor life expectancies in the aerospace, medical, and transportation industries, to name a few. Weibull described and modeled the phenomena in 1939, his distribution has given good service in modeling capacitors, bias temp inversion of transistors, incandescent light bulb failures, and many other systems since. And all of the major capacitor manufacturers use it to guide the customer through the life expectancy calculations of the product in their application notes. In fact, of the capacitor types, electrolytic caps are about as close to bathtub as any. This was way outside of the scope of this thread which is a review of a preamp, but since it was asked, here are just a couple of manufacturer's application notes with detailed discussions of the bathtub curve:

https://www.cde.com/resources/technical-papers/AEappGuide.pdf

https://www.nichicon.co.jp/english/products/pdf/aluminum.pdf

This thread is a measurement vintage gear measurement, not a blanket replacement of capacitor discussion, or a reliability discussion. I regret it keeps getting drug into the weeds. The capacitors aren't this preamplifier's problem anyway. It does have a couple of issues. One is the performance of the volume control that I mentioned in the first post.

In fact the preamp used first-class caps for the time, operates them well below the voltage rating, and most importantly operates the caps at just above ambient temperature. Which if we follow the rel estimates provided in the above links, you will see the lifetime derate due to the low operating temp gives wear-out estimates of decades. And if the are degrading, it isn't some catastrophic issue like many other component.

Aluminum electrolytic infant mortality rates are actually quite high, and represent a greater risk to rel than an aged capacitor so long as it is operated below it's max voltage, and especially max temperature. Critical military, transportation, and medical applications specify multiple layers of acceptance testing to prevent components like new capacitors from introducing infant mortality fails into freshly repaired systems. This is just a piece of audio equipment so the stakes are really quite low, and if a cap goes bad in the next decade it isn't the end of the world, and the caps are not degrading the sound or operation in any way shape or form.

So many just think that ONLY capacitors are an issue. There is wire connection corrosion, relays, and many other things. Capacitors are just a small part of issues that occur on 35+ year old gear. How it was kept, (salt air environment, heavy smokers causing a coating to help overheat everything, High pet dander, set in a hot, humid attic for years, you just don't know. (Unless you bought it new [or a close friend did] you ABSOLUTELY do not know). I recommend getting it tested and having it gone through. It may or may not need caps and may or may not need lots of things or may not anything at all.
I never meant what I said to be only relating to one type of part.
Now that I have clarified what I meant, perhaps we can get back to the discussion of the OP.

 

[BigVU's]

Did some one say VU meters? The OP's set up looks juicy for 2025 and beyond! Thanks for posting this.