Well that's a load of information, thanks
Got one channel out of the shed and the power supply (wrong one with the MPSA42/92 instead of the SB716/SD756) to start playing with it.
First replace the MPSA with the original SB/SD pair
Second replace 100Ohm gate stoppers with 220Ohms and play with R3 (47K) and go from there
Got one channel out of the shed and the power supply (wrong one with the MPSA42/92 instead of the SB716/SD756) to start playing with it.
First replace the MPSA with the original SB/SD pair
Second replace 100Ohm gate stoppers with 220Ohms and play with R3 (47K) and go from there
LTspice doesn't accurately simulate these VHF oscillations (they are affected by layout for instance) - err on the side of larger gate-stopper resistors.
Maybe the miller cap is too small. Use 68p or 100p instead.
I don’t think the gate stopper is that important.
I don’t think the gate stopper is that important.
I found that 200 Ohms was sufficient to stop the oscillations, when I was experimenting with different values.
I built two with 200 Ohms, and two with 220 Ohms.
Both have worked fine.
MOSFETs don't care what you think, they just do what they do.
Given the right conditions to oscillate, then they will oscillate.
Last edited:
I should clarify my observation: Higher gate stopper values had no discernable effect on the audio signal 🙂
It depends what mosfet you use. For ecw20n20, it doesn’t matter, because its “Reverse Transfer Capacitance” (16p) is so small. The time constant of the RC network it creates is so tiny that it affects almost nothing regarding audio signal.
It will be another story if you use some switch type mosfet with 500p Reverse Transfer Capacitance.
Last edited:
I was referring to my LTspice simulations of the Maplin MOSFET amplifier, the stopper resistors could get quite a bit bigger without degredation of the audio waveform 🙂
But thanks for the note about reverse transfer capacitance, I'm guessing this type need a low impedance drive?
I had a glance on the specs of 2sk135. They are very similar with modern counterparts such as ecx10n20. Your observation is correct.
Ok guys, sorry for the delay had a very heavy week.
For information:
Mosfets and transistors are the ones called on the article so 2Sj50/2SK135, 2SA872A for the input dif pair and 2SB716/2SD756A.
Power supply is the test one with 35/0/35 Vac at 300VA that gives 48/0/48Vdc at the capacitor bank (4x10000uf).
So just tried all (most) of the suggestion above and ended up with 240 Ohms gate stoppers and 51K at R3 position (was 47K) without much change just had to trim the idle current to 120mA (dropped to 95ma)
No heating from any of the transistors all nice and cold
Idle current varies around 2mA but I noticed that the input voltage does the same (grid variations ?) 48 up to 49.something.
Had it playing for around 40 minutes and all seems fine.
Suggestions ?
For information:
Mosfets and transistors are the ones called on the article so 2Sj50/2SK135, 2SA872A for the input dif pair and 2SB716/2SD756A.
Power supply is the test one with 35/0/35 Vac at 300VA that gives 48/0/48Vdc at the capacitor bank (4x10000uf).
So just tried all (most) of the suggestion above and ended up with 240 Ohms gate stoppers and 51K at R3 position (was 47K) without much change just had to trim the idle current to 120mA (dropped to 95ma)
No heating from any of the transistors all nice and cold
Idle current varies around 2mA but I noticed that the input voltage does the same (grid variations ?) 48 up to 49.something.
Had it playing for around 40 minutes and all seems fine.
Suggestions ?
Sounds good!
The MOSFET amplifier is designed to run at between 50mA and 100mA (111mA total IIRC) idle, but the MOSFETs need to be reasonably heat-sunk to keep them coolish - (2 x 50)V at 100mA = 10W.
Do you mean gate, rather than grid?
Where are you measuring input voltage?
There is a DC servo effect (via the GNFB) that may shift the voltage of the bases in the input LTP around slightly as it works.
If you have the energy I can recommend my mod of a CCS on the input LTP (a couple of pages back), but I think this also necessitates a few slight value changes and an additional resistor. This isolates it from the supply rail noise and lowers distortion anyway.
Hi
I meant the grid, I did measure the voltage on the transformer leads and it varies ever so slightly but on the other hand I was working on the kitchen table and the wife had the oven on and maybe that caused the voltage to change from 48 to over 49Vdc.
All other measurements were made at the board terminals.
I don't mind your ccs mod but I would like to keep it at original as possible, going to finish the 2nd channel and see how it goes.
I have here another article with a class A amp that uses the same output pair (2 x j50/2 xk135 ) with 2A of idle current at a power rail of 20Vdc maybe I give that one a go but right know I'm in the early stages of building something with 5 pairs of the above Mosfets.
PS: I think I found my 2SD756 heating issue, somehow I forgot to change my temporary diff input pair for the 2SA872 and still had the BC640 in the board.
Will find out during the next week.
I meant the grid, I did measure the voltage on the transformer leads and it varies ever so slightly but on the other hand I was working on the kitchen table and the wife had the oven on and maybe that caused the voltage to change from 48 to over 49Vdc.
All other measurements were made at the board terminals.
I don't mind your ccs mod but I would like to keep it at original as possible, going to finish the 2nd channel and see how it goes.
I have here another article with a class A amp that uses the same output pair (2 x j50/2 xk135 ) with 2A of idle current at a power rail of 20Vdc maybe I give that one a go but right know I'm in the early stages of building something with 5 pairs of the above Mosfets.
PS: I think I found my 2SD756 heating issue, somehow I forgot to change my temporary diff input pair for the 2SA872 and still had the BC640 in the board.
Will find out during the next week.
Ahh - the 'National Grid', haha, yes, I get it now!
I hope the roast is good 😀
Have you got a box and layout in mind?
I bypassed my PSU caps with some polyesters (as I had some spare from a speaker recap, where I had to order about 10), and in both my amplifiers (I have two pairs, one in monoblocks, the other with the CCS mod in an old Sony F6B) I isolated the driver part from the main PSU.
In the last version - the CCS/F6B I used resistor-schottky and bigger caps to isolate it. Then sudden dips in the rail don't get passed on to the driver stage.
In the early 1980s when I was soldering my first pair of these together, I had no idea they'd still be 'a thing' for so long 🙂
3 things
The roast chicken was lovely thanks
The power supply I got off *bay and has schottky diodes 4 big Nippon chemi caps with 100nf poly. ( the last one I bought came with fake Hitachi caps)
The box/case I normally build my own but I have to change the heatsinks the ones attached to the Maplin module is "only" 300x200x60 so just a big overkill.
They don't build parts specially power transistors/mosfets like they used to everything is been discontinued or changed to SMD.
Done for today have to get out of bed at 5.30
Have a good one
Ricardo
The roast chicken was lovely thanks
The power supply I got off *bay and has schottky diodes 4 big Nippon chemi caps with 100nf poly. ( the last one I bought came with fake Hitachi caps)
The box/case I normally build my own but I have to change the heatsinks the ones attached to the Maplin module is "only" 300x200x60 so just a big overkill.
They don't build parts specially power transistors/mosfets like they used to everything is been discontinued or changed to SMD.
Done for today have to get out of bed at 5.30
Have a good one
Ricardo
😀 Bon Appetite!
Now on eBay I take great care to select the seller, or from China get Chinese branded caps, as I figure that branded chinese are better than faked caps 😀
Re. the Schottky diodes, I'm not talking about the main PSU 🙂
What I did was cut the traces between the output mosfet drains and the PSU on the PCB.
So the main PSU power goes, via inline fuses, to the output follower MOSFET drains. Dirty power there doesn't worry me so much, so PSU sag will and does occur there.
But the driver stage, home of the control loop and small signals, is powered via schottky diodes (for low Vfwd voltage drop, and to prevent back-feeding the MOSFETs), and 39R (IIRC) resistors to soften the recharge of the on PCB rail caps. I'll lose < 1V doing this, but the control section power rails no longer dance withe the music so much.
On my monoblock version of these boards, I also cut the tracks, but used a seperate bridge and caps for the driver section, which is better still. But that one hasn't got the CCS mod etc, so it's not all perfect 🙂
Of course it really needed new PCBs, so I could lay it out properly, but budget and time meant a small bodge with the heatshrink instead LOL. Also if I made a new PCB, I'd want a tightly regulated supply of about +60V on the driver board, with perhaps some studying of the ground current path. Then I'd get 'feature creep' and end up with a new amplifier - haha. 😀
Attachments
Hi everybody
Got the 2nd channel working and set
I completely forgot that I still had the BC640 diff pair in the board and it seems that those transistors were the cause of the heating issue on the 2SD756.
So to recap had the wrong input pair causing heating issues on TR4 (2SD756) that I replaced with 2SB649 to see if I could bring the temp down without success.
Now with the right transistor everything is back to normal without overheating of any of the transistors (even with all original resistor values).
Up to the next project
Thanks to everybody that chimed in
Ric
Got the 2nd channel working and set
I completely forgot that I still had the BC640 diff pair in the board and it seems that those transistors were the cause of the heating issue on the 2SD756.
So to recap had the wrong input pair causing heating issues on TR4 (2SD756) that I replaced with 2SB649 to see if I could bring the temp down without success.
Now with the right transistor everything is back to normal without overheating of any of the transistors (even with all original resistor values).
Up to the next project
Thanks to everybody that chimed in
Ric
