I have built an amp using two NAP140 ebay kits. It has separate power supplies for each board, it sets up nicely with low offset (laess than 14 mv) and the static bias is set to 38ma. One channel works perfectly but the other when run at moderate level heats up more than the other. When I remove the signal source and short the input to ground the current increases but the offset stays low to the point where the transistors are very hot. Switch it off and let it cool down and it is normal again until driven once more, any ideas?
microx
microx
Yes, I thought that but it still does it with the input shorted to ground, anyway there is no problem until it is driven but I will scope the output to check, thanks.
microx
microx
Justamente, puede ser que el ampli sea "condicionalmente estable" y sea la señal de entrada que lo haga entrar en oscilación, y después quede automanteniéndose oscilando.
si, puede ser que con señal hay mas calor y este causa que empieza a ser inestable, si no siempre hay
No, me refiero a que cuando lo encendés, no oscile de por sí, pero al aplicarle señal de audio, esa señal arranca o detona la oscilación, y una vez que empieza la misma, ya se autoalimenta y no cesa hasta que lo apagás. Eso supongo y entiendo a partir de tu primer post.
Salu2.
Guys, this is an English language forum. Please see the forum rules.update
well I cant see anything on the scope and I am starting to suspect the output transistors (2SC3858) they are marked Sanken but who knows. Farnell have no stock but the chinese have tons, makes you think. Years ago I had some 2N3055 that did something similar, I replaced them with known good ones and cured the problem. Cutting the top off revealed something very odd and nothing like what was inside a genuine item.
microx
well I cant see anything on the scope and I am starting to suspect the output transistors (2SC3858) they are marked Sanken but who knows. Farnell have no stock but the chinese have tons, makes you think. Years ago I had some 2N3055 that did something similar, I replaced them with known good ones and cured the problem. Cutting the top off revealed something very odd and nothing like what was inside a genuine item.
microx
All the same microx, you have 1working channel. Don't give up if you don't see oscillation - assuming your scope is at least 20MHz bandwidth. It can be hard to see and mosfets often oscillate above this range (you may not see them on anything less than 100MHz scopes)
Try something simple first - reset the bias to 10mA and come back with the result. You could be right about thermal instability. After all, there is no thermal coupling to the output transistors and the original design works on the ambient case temp. - a very slow "feedback" system.
IME these designs need to be in a case and mounted at the top, like the NAP 140, to get reasonable warm-up time. Testing modules on the bench can be dangerous, as you may have found, since there is even less thermal control over bias. 38mA is absolute max. and most recommend no more than 30mA bias so don't worry about cutting bias. I believe that some models had negligible bias with the same topology.
Try something simple first - reset the bias to 10mA and come back with the result. You could be right about thermal instability. After all, there is no thermal coupling to the output transistors and the original design works on the ambient case temp. - a very slow "feedback" system.
IME these designs need to be in a case and mounted at the top, like the NAP 140, to get reasonable warm-up time. Testing modules on the bench can be dangerous, as you may have found, since there is even less thermal control over bias. 38mA is absolute max. and most recommend no more than 30mA bias so don't worry about cutting bias. I believe that some models had negligible bias with the same topology.
Is the NAP 140 clone a quasi topology?
If so then the required bias for minimum crossover distortion could be very much less than 38mA.
I wonder how that converts to Vre?
If so then the required bias for minimum crossover distortion could be very much less than 38mA.
I wonder how that converts to Vre?
another update
I have tried setting the bias to just a few m/a with the same result, everything is mounted in a case and I let it idle for two hours before setting the bias. My scope is good for 40 mhz but is see only the normal "grass" all through the signal path. Now I have had a burn up taking out most semiconductors on the plus voltage side. I think it will be cheaper to buy another kit than source components. I think that this points to crappy semiconductors, the vexing part is how well the other channel works! I did it all by the book, winding it up slowly on a variac with the bias at minimum etc. Never mind, failures sort the men from the boys, I once built an underwater flashgun that caught fire underwater!
microx
I have tried setting the bias to just a few m/a with the same result, everything is mounted in a case and I let it idle for two hours before setting the bias. My scope is good for 40 mhz but is see only the normal "grass" all through the signal path. Now I have had a burn up taking out most semiconductors on the plus voltage side. I think it will be cheaper to buy another kit than source components. I think that this points to crappy semiconductors, the vexing part is how well the other channel works! I did it all by the book, winding it up slowly on a variac with the bias at minimum etc. Never mind, failures sort the men from the boys, I once built an underwater flashgun that caught fire underwater!
microx
Quasi
Yes - and near original apart from semis and VI limiters. Higher figures for bias have been promoted for years on Neil Mcbride's site and the Pink one. I've used slightly higher bias on clones and my original NAP140 came from the local service agent many years ago with 27 and 30 mA settings. I understand this was determined by actual THD measurement, not a particular current and led to quite a range of settings.
Vre on the CFP side is always a puzzle, but old CFP stages have strange bias systems anyway. Hitachi for example, used high, 100 mA bias to get stable control with the bias circuit they once used. They sensed the output heatsink which is not ideal. At least, you cannot easily hold a "correct" level ~13 mA, by sensing the outputs.
P.S. sorry to hear that, microx. I would replace outputs with 2SC5200 which are cheap and used to be fine before these fakes were supplied, The other semis were probably fine.
Yes - and near original apart from semis and VI limiters. Higher figures for bias have been promoted for years on Neil Mcbride's site and the Pink one. I've used slightly higher bias on clones and my original NAP140 came from the local service agent many years ago with 27 and 30 mA settings. I understand this was determined by actual THD measurement, not a particular current and led to quite a range of settings.
Vre on the CFP side is always a puzzle, but old CFP stages have strange bias systems anyway. Hitachi for example, used high, 100 mA bias to get stable control with the bias circuit they once used. They sensed the output heatsink which is not ideal. At least, you cannot easily hold a "correct" level ~13 mA, by sensing the outputs.
P.S. sorry to hear that, microx. I would replace outputs with 2SC5200 which are cheap and used to be fine before these fakes were supplied, The other semis were probably fine.
Last edited:
limiters
the clones I have retain the limiters (some others dont). Some advocate removing them citing increased distortion, I guess as the limiter gets on it's "knee". I have ordered another pair so I will end, I hope!, with three working boards. I think I will loop one through RMMA and measure the THD with a range of bias settings and with the limiters removed and with other O/P semis and implement my findings on the other two boards.
microx
PS thanks for your interest
the clones I have retain the limiters (some others dont). Some advocate removing them citing increased distortion, I guess as the limiter gets on it's "knee". I have ordered another pair so I will end, I hope!, with three working boards. I think I will loop one through RMMA and measure the THD with a range of bias settings and with the limiters removed and with other O/P semis and implement my findings on the other two boards.
microx
PS thanks for your interest
Thanks also for sharing, microx. I don't recall seeing these clones with limiters but would you have a link or sellers name for reference? Your view on limiters is shared by many here who believe they are not good for high quality audio
However, a designer has to protect his product and before output relays were common, I guess this was the way to go. At low power, I don't believe there is any audible effect but some golden ears can hear the brand of resistors, too. 🙄
However, a designer has to protect his product and before output relays were common, I guess this was the way to go. At low power, I don't believe there is any audible effect but some golden ears can hear the brand of resistors, too. 🙄
ebay item number
I bought mine from "anitaydavid" item number 320885371709 check him out. I have had a few things from him and he seems OK. I cant believe you are so sceptical about resistor brands (it's the paint for the colour bands that counts). If you like I can put you in touch with a man who will sell you ultra low distortion volume knobs made from dinosaurs bone and doped with mermaids lanolin. You can tell that to the marines! LOL.
Yes, I think relays do a better job than limiters.
microx
I bought mine from "anitaydavid" item number 320885371709 check him out. I have had a few things from him and he seems OK. I cant believe you are so sceptical about resistor brands (it's the paint for the colour bands that counts). If you like I can put you in touch with a man who will sell you ultra low distortion volume knobs made from dinosaurs bone and doped with mermaids lanolin. You can tell that to the marines! LOL.
Yes, I think relays do a better job than limiters.
microx
(Sigh) I love a good laugh! Thanks, these are LJM (member ljm-ljm) boards like those from Tubeshunter, Along1986090 and others too. 'Sorry I missed seeing that they do have limiter circuits.
A note to those concerned about the parts: Tubeshunter now offers the option of matched 2SC5200(O) outputs which should be a safer bet. After all, 2SC3858 has been obsolete for many years and these ones look consistently brand new.
A note to those concerned about the parts: Tubeshunter now offers the option of matched 2SC5200(O) outputs which should be a safer bet. After all, 2SC3858 has been obsolete for many years and these ones look consistently brand new.
An afterthought for anyone neediing a credibiliy check. Note: the model is different but the design very similar, like all Naim originals
Naim, NAPs, Naits, distortion and bias-current settings
Naim, NAPs, Naits, distortion and bias-current settings
- Status
- Not open for further replies.