Luxman L-85v Integrated Amplifier

[pma]

With 2 pairs of MJL21194/3 (instead of 1 pair), there should be a substantial rise of output power. 260W/4ohm, and THD+N (BW 20kHz) goes where it should be.

 

[sabristol]

Nice work @pma. I’m tempted although practically more heat to dissipate and physically difficult to fit them in. Is that where you’re going?

 

[pma]

The case would look like this, so no problems with heat dissipation

http://pmacura.cz/pa2.htm

BTW, I can see you made good changes to the original design, which shows the dust of ages

and was probably not very reliable, due to typical undersized construction and 1 pair of not very robust output devices. As mentioned in

https://www.vinylengine.com/turntab...p=684929&sid=a9dd5f3f6891bac3a9a913b74a004688

However the basic idea is good and may be improved, as you did.

 

[pma]

Hi @sabristol ,

as you brought the schematics in here, it would be fair for me to display what I have drawn. Attached please find the current status of the circuit that I want to make as a functional sample.

 

[sabristol]

Hi @pma

This is excellent. What you've done is exactly what I wanted to happen when I posted it - i.e. someone take it, model it, suggest improvements based on analysis (not subjectivism) which is in keeping with the ethos of this site. Spot on.

Thanks for sharing back, it's appreciated. For me it's just me trying to breath some life into a childhood hobby anyway so I don't feel protective over the work I've done. I just want to catchup lost time and learn something.

My circuit separated out the signal-gnd from the power-gnd, whereas you've merged. Will it be ok? I guess it depends on your overall earthing arrangement. There would be hum in the Luxman.

I like the addition of the DC servo. I've not got any DC offset problem so left as is.

Re your previous post using two pairs of o/p bjt, I'd be tempted to move to SMT if I do another version, so I could squeeze more on including the DC servo....now look what you've done

 

[sabristol]

The case would look like this, so no problems with heat dissipation

http://pmacura.cz/pa2.htm

BTW, I can see you made good changes to the original design, which shows the dust of ages

View attachment 117270

and was probably not very reliable, due to typical undersized construction and 1 pair of not very robust output devices. As mentioned in

https://www.vinylengine.com/turntab...p=684929&sid=a9dd5f3f6891bac3a9a913b74a004688

However the basic idea is good and may be improved, as you did.

The OEM schematic diagram I know it well! A few cost cutting measures. As mentioned earlier, Z501 / C504 are not grounded there, they go to the 0V, and D503 is 4k7, not 5k6.

I previously misunderstood, you are not squeezing the doubled up o/p transistors into the L-85v case and onto the existing heatsink, although I'm now wondering if it's feasible....??

 

[pma]

The OEM schematic diagram I know it well! A few cost cutting measures. As mentioned earlier, Z501 / C504 are not grounded there, they go to the 0V, and D503 is 4k7, not 5k6.

I previously misunderstood, you are not squeezing the doubled up o/p transistors into the L-85v case and onto the existing heatsink, although I'm now wondering if it's feasible....??

Hi, I will put this circuit into my prototype case, which looks like this:

I make the PCB boards same size and drills as the boards in the photo, to have the effort minimized. The PSU voltage is 2x56V, so it is perfect for this project.

 

[sabristol]

That's a cracking job! I'm impressed.

It will be great to see test results in due course. What sort of timeline are you on?

Four of those TO-264 must be easier to fit into the L-85v.

 

[pma]

What sort of timeline are you on?

Short. PCB's just ordered

 

[pma]

There seems to be a big issue with appropriate VAS transistors worldwide availability for this project (A1381/C3503 or similar). Unobtainable or lead time 12 weeks or so. They must be fast (fT = 150MHz) and high Vceo (150V). In case of slow parts the circuit would happily oscillate or the frequency compensation would kill the parameters.

So what now. BD139/140 are fast enough but Vceo only 80V. So I made a compromised decision, I will use 2N5401/5551 and VAS current reduced to 3mA to stay safe in Ptot.

The world becomes more and more narrow (because of parts disappearing) for analog designers who refuse to admit to be a mere assemblers of OEM class D modules only. Buy a module and wire it in the box = no fun.

 

[sabristol]

@pma don't give up on BJT!!

Mouser can dispatch immediately 9,292 of the KSA1381ESTU, and although KSC3503 seems to be a problem at Mouser, Farnell UK have 2544 in stock.

 

[pma]

@pma don't give up on BJT!!

Mouser can dispatch immediately 9,292 of the KSA1381ESTU, and although KSC3503 seems to be a problem at Mouser, Farnell UK have 2544 in stock.

Thank you, yes there are some possibilities, e.g. RS components would send me them immediately, but 60 + 60 pcs. I went for a compromise, Zetex ZTX855+ZTX753, immediately.

BTW the frequency compensations have to be modified in any case, because R24+C13 (as per your drawing) restrict +slew rate too much, making it much lower than -SR.

 

[sabristol]

@pma I took a look using my LTspice model and couldn't see a difference between +SR and -SR which measured ~230 V/ms. Probably me not knowing what I'm doing.

 

[pma]

I do not think you do not know what you are doing. It is rather a difference in some components between your and my circuit, and also a different simulator thus models. Dynamic parameters, however, have to be verified on a real circuit. So you may measure +SR and -SR with an oscilloscope, if you like. This is to be measured without input RC filter.

 

[pma]

I took a look using my LTspice model and couldn't see a difference between +SR and -SR which measured ~230 V/ms.

I have just re-read your post. 230V/ms is 0.23V/us and this would be extremely low and unusable slew rate for the power amplifier. Are you sure you state 230 Volts per millisecond?? Just FYI, 10kHz wave with 10V amplitude has slew rate of 600V/ms (0.6V/us). Did you disconnect the C8 capacitor for SR simulation?

SR = dv/dt(max) = 2*pi*Vp*F

 

[sabristol]

@pma - ah yes, accepted, thanks. My LTSpice simulation was nonsense, even the input pulse has the same slew rate! As I suspected, more LTspice user training required.

(btw I noticed some of the component values in my model weren't the same as the actual. They were a carryover from me experimenting previously)

 

[pma]

@pma - ah yes, accepted, thanks. My LTSpice simulation was nonsense, even the input pulse has the same slew rate! As I suspected, more LTspice user training required.

(btw I noticed some of the component values in my model weren't the same as the actual. They were a carryover from me experimenting previously)

You most probably need to set time step in your time analysis to 1us or less, like 100ns or 10ns (see below). Edit> and also the square wave generator time step.
I also have some changes in my circuit, this is the current version. Slew rate is 22V/us. Please take into account that the capacitor in the input RC filter must be zeroed for the SR investigation.

LTSpice setting example

 

[restorer-john]

So what now. BD139/140 are fast enough but Vceo only 80V.

What about MJE-340/350 or BF-469/470 pairs. There should be tons of them in Europe.

 

[pma]

What about MJE-340/350 or BF-469/470 pairs. There should be tons of them in Europe.

Speed is important. There is an EF before VAS, added phase shift must be thus minimized for stability. Loopgain analysis was made, this is crucial. Zetex are already on the way and will arrive on Monday to my place, so this is fixed. PCB's to arrive in the middle of the week.

 

[restorer-john]

Speed is important.

They are FT 10MHz for the MJE and 60MHz for the BF-469/70 with 250V VCE. The BD-139/140 is 50Mhz.

I wouldn't use a Zetex transistor for anything- I just don't trust them. Back when Ferranti was Ferranti, but then they got sold out, restructured, sold again etc.

Anyway, just trying to help Pavel.