bias a trace elliot ta100r

[lowell]

They will work but they are in a completely different package, which could cause some problems when trying to use them. They also require slightly higher voltage on gates to work properly (so same small changes to the amp may be required). If you are looking for better equivalents, you may take a look at BUZ901/906. I also see (quite often) 2SK135 on ebay. In Poland I can even buy them in several shops - hard to believe that you cannot buy them in the States.

I checked for them and they seemed quit expensive. In any case I've had these replacements installed already, on the heatsink, with thermal pads and wires running to the board. I'll look into the BUZ901/906. I think I looked for those already and couldnt' find any either but will check again.

In any case at this point I have the 1.5k 10w resistors in and there is STILL 70mv on the output.

[MarkusBass]

Take a look here:

BUZ901 - SEMELAB - N CHANNEL MOSFET, 200V, 8A, TO | Newark Products

BUZ906 - SEMELAB - P CHANNEL MOSFET, -200V, 8A, TO | Newark

Mark

[52 Bill]

Thanks for the post Mark. If you look at the data sheet, it's all in Russian maybe?

[MarkusBass]

Thanks for the post Mark. If you look at the data sheet, it's all in Russian maybe?

This is possible (I cannot check it today - I have a computer without Acrobat Reader). I think you can find on the net some information about the factory in Russia. This may be the same story as with valves - every good valve was manufactured in Russia. The Russian language is not that difficult, especially in a datasheet there is a lot of symbols, or drawings that you can easily recognise.
If you want the datasheet in English, you may go to Profusionplc: Lateral Mosfet
I think that some of the transistors listed there are equivalents to the ones that Lowell is searching for.
You may also ask Km6xz to purchase aa batch of BUZ9xx - they must be cheap there (and he stays in Petersburg - most probably close to the factory).

Mark

[J M Fahey]

I think they linked to the Russian Distributor page by mistake.
Semelab/Magnatec is British.
Their proper page is:
TT electronics Semelab Limited - Magnatec Datasheets

[lowell]

Well at this point I have the 1.5k 10w resistors in there and no power mosfets. I am very confused at results I'm getting.

I have ZD3, ZD4, ZD5, ZD6 all lifted
There is -40mv on the output (no speakers connected)
On D7 anode and D10 cathode I have -51v (I have P1 set at zero ohms)

I began checking voltage drops across R23, R21, R22
I got ~500mv across R23 and thought this was half of what it SHOULD be. I checked voltage drops across R26, D6, R22 string and it all looked good, therefore TR5 is most likely fine - so I replaced TR6. I now have 1v across R23. However, still -51v at P1.

So I'm now questioning if TR4 is working. I am getting 590mv across R21 and Vb/e on TR4 is -560mv. So I'm thinking that it is fine. I also lifted C17 and it didn't help.

So everything is checking out fine, BUT, -51v on the junction of TR4 and TR6. I'm very confused.

[Steve Conner]

That's right, the Semelab parts (or at least the silicon chips inside them) are made right here in Scotland. Exicon and all of the other little brands use the same chips, so they are equivalent.

The only other people still making lateral MOSFETs are Hitachi, the 2SK1058/2SJ162 etc. These aren't drop-in replacements due to a different threshold voltage.

[lowell]

*bump*

[lowell]

So I have the resistors in place of the mosfets. I also have 22k 1/2w resistors in place of the drivers. I still had a bit of offset at P1 when it was set to zero resistance. I then decided that I'd continue this "lets see what results we get as we eliminate variables." So I disconnected the 56k feedback resistor and this modification allowed for 0v offset at P1. SOooo... I then checked that TR5 was on and drawing current.. and it was. The B/E junction was ~ 600mv as well. Finish this post in a bit.

[MarkusBass]

I remember my advise to put resistors in place of MOSFETs. But I don't know who advised you to put 22k resistors in place of drivers (do you mean TR4, TR6?). They, togehter with TR5, form a diffenrential amp with a current source. So you removed the current source and made it "non-differential" and now you complain that it does not work. Of course you may say that a high value resistor is kind of a current source but it is not the way to fix the amp. You are now triyng to solve problems that you created and not the real one.
This amp is a very simple amp and if you have such problems with it, you could replace all semiconductors (I mean 5 transistors and 8-9 diodes in the output stage).

PS:the problem with 40mV on the output may be caused by rails (uneven voltages), or uneven 1.5k resistors. Have you considered this?

Mark

[lowell]

Mark I stated that I'd finish my post later. As I had to tend to something. I'm merely trying to achieve symmetry on P1. Which I did when replaced TR4/6 with 22k resistors AND disconnected the 56k feedback resistor. I decided on 22k as those are what I calculated would provide me with ~500ma of current. After this I reinstalled TR4/6 and am back to having bias problems. Both sides of P1 are close to the negative rail at -53v. I then thought maybe I damaged TR4/6 in this process. Maybe TR4 is open or TR6 is shorted. Nope, still the same issue after replacing them again. I have checked and double checked that all connections with my meter, from lead to lead, are indeed connected. I cannot for the life of me figure out why TR4 collector is at the negative rail ( and P1, and TR6 collector). Of note is that I removed C18 thinking maybe it was leaking and pulling TR4's base down and turning TR4 on harder, but no change in results here. Here are voltages on TR4/6

TR4
B/E: -560mv
TR6
B/E: +600mv
C/E: 160mv

R21 has 520mv across it - thus TR4 is on and drawing current.
R23 has 1v across it, which makes sense if both TR5/6 are pulling current through it, so TR6 is on too.
R22 has 460mv across it so TR5 is on too.
TR5
B/E: 600mv

I'm quite confused as everything seems good with the meter, however P1 is at the negative rail!! ??

[lowell]

Anyone have any thoughts on this? Considering replacing all PA devices at this point.

[lowell]

Ok so back at this again. The 1.5k resistors are in place of the MOSFETS. I currently have -56.3 volts at TR6:C AND TR4:C. Again I've checked continuity from lead to lead all over this board. All is good. Here are some other voltages worth noting:

P1 is set to 0ohms
RAILS: +/- 56.3v
D6 cathode: 55.2
568mv across R21
1.1v across R23
1.76v across R16
1.64v across R18
TR5:C 20.6v
TR5/6:E -55.2v
TR6:BE 676mv

There should be ~0v at P1, not -56.3v

I disconnected TR6:C just to make sure there wasn't some shorted connection on the board. When I did this the voltage at P1 flipped to the positive rail. So my hypothesis is that TR6 is saturated. I don't THINK that a Vbe on TR6 of 676mv would do this. As an experiment I placed a 50k pot across R16 and tried turning it ever so slowly with my meter (in attempt to adjust TR6 base voltage) measuring voltage on P1. There is a point on the pot where the voltage flips back and forth between negative and positive rails. I've already replaced TR6 at least once if not twice, but I'm considering doing it again. It seems TR6 cannot be biased in it's "operating region."

Thoughts??

[Enzo]

I just scanned through all this from the start. Not exactly sure where we stand. Are we still wondering why the bias control doesn't change anything? Did we power off and measure the trimmer resistance to see that it still works as a pot? An openb wiper on the trimmer and you have a 220 ohm resistor sitting there.

Also, way back, you adjusted it for crossover notch and it worked? Why did we keep fixing it after that point?

COmment on terminology. As a rule "power supply" refers to the internal circuits of an amp, not the mains. Rectifiers, filter caps, regulators, etc. We do not geneerally refer to the mains as "power supply". Yes, the mains are ultimately the source of power for the circuit, but we don;t call it the supply.

[lowell]

Enzo,
yes indeed still wondering why the bias VOLTAGE isn't right. The bias pot works, it goes from 0-220ohms when adjusted. The collector voltages of those surrounding transistors TR4 and TR6 however are not correct. At this point, the bias pot is set at 0ohms and there is -56.3 (negative rail) on it. I think this point should be ~0v.

Now that this is cleared up, you can review my last post with all those listed voltages etc and it may make more sense.

[Enzo]

The bias control won't affect the voltage it sits at, it controls the space between TR4C and TR6C. It ought to rest at about zero, but it moves as it is told.

SO the whole middle is sitting at -50. Yoy have +50 on the emitter TR4, and +20 on its base, yet its collector is pulled all the way down to the bottom rail. Looks like TR4 is not turning on.

[Steve Conner]

If the feedback loop is broken, it's normal for a SS power amp to flip to one of the rails. If you removed the output devices, you have to close the feedback loop some other way.

A while back I designed a hi-fi amp, and decided to do it in two sections like the old Peavey PA amps. I made it so that the driver board could be tested on its own by hooking its two output terminals together and jumping them back to the feedback input. It was totally incapable of driving any load, but if it was working properly, the DC offset, gain and maximum output level would be the same as if it were hooked up to the output stage. That was a big help in troubleshooting. Maybe you can do something similar to your thing.

[lowell]

Steve I have reconnected everything. The amp is wired like the schematic, except for the MOSFETS. I still have the 1.5k in their place. Is there no feedback loop with it setup like this? And by feedback loop you are referring to R14 56k right?

Of note, now that everything is back in place, is that I have -12.5v at P1. This is interesting because it's clearly 12.5v due to the diodes D10/ZD5. Being that there is ~0v on the output this only makes sense. But now, why is there negative voltage like this on P1? Again, without D10/ZD5, P1 has the negative rail voltage. And Enzo, TR4 seems on as there is 620mv across R21.

All those diodes between the upper and lower half of the PA check out good with the meter, however they're the ONLY semiconductors that I haven't tried replacing so I'm tempted to just do it and see.

Finally, clearly I don't REALLY understand this circuit and need to do some reading. What is up w/ R15? Fyi it has 0v across it maybe that is right, or maybe not. Can someone direct me to reading the theory behind this specific PA?

[Enzo]

I may be steering you wrong, Lowell, but you have current flowing into TR4, but it could all be leaving via the base. If the base current can;t control the collector current, the transistor is defective.


Feedback. The output is driven by - in this amp - th MOSFETs. The amp reacts to changes on the output and tries to correct them. If the circuit cannot drive the output bus, it gets no feedback information. And without the MOSFETs, it cannot drive the output. If the MOSFETs are not there, the amp can;t work right.


Why is there negative voltage at P1? Because TR4 cannot - for whatever reason - pull it up positive to get it to zero. TR4 could be defective, or it coulod be a problem with the rest of the circuit unable to control it.

You have a series circuit: +50v, R21, TR4, P1, TR6, R23, -50v. P1 OUGHT to be centered in the middle of that, right around zero. Current flows through that whole string, so to get any one point to be more positive than it is, you need the things above it to conduct harder.

R15 (and R14) are feedback. R14 is direct from the output bus. R15 monitors the gate drive. Either end of P1 is the drive signal to a MOSFET gate. If the drive signal voltage goes high enough - in this case about 34v - to turn on D8 and ZD4 (or the equivalent on the negative side) Then it sends that excess voltage through R15 to TR3 to tell it to stop. SO this feedback path acts as a limiter. So until you get that much voltage on P1, no current will flow through R15, hence no voltage drop across it.

[lowell]

"Feedback. The output is driven by - in this amp - th MOSFETs. The amp reacts to changes on the output and tries to correct them. If the circuit cannot drive the output bus, it gets no feedback information. And without the MOSFETs, it cannot drive the output. If the MOSFETs are not there, the amp can;t work right."

I'm confused now... so do you think I should put the MOSFETS back in? Or try to fix the bias first without them so as to not damage them? Say the MOSFETS were working right, then we'd have ~0v on the output, just as we do now (well it's 180mv actually). So what's the difference here being there's no signal flowing at this point in the repair? As I understood it, the 1.5k resistors are to simiulate functioning, WORKING, MOSFETS, so as to try and narrow down the bias issue. Finally, I've replaced TR4 once, if not twice. I'll do it again right now to be sure.

[Enzo]

Well if you already replaced it, it may be fine. Unless the new one was damaged by the circuit at some point.


If you have resistors, and the output winds up off just a hair, the differential circuit will move to try to correct that. But of course that results in no change, because we have resistors instead of active transistors. SO the differential tries even harder, then harder, until it is slapped over to a rail as Steve mentioned.

Ever go into a skid on ice in your car, locking up the wheels? You push on the brakes to slow down, but the car keeps going - doesn't slow down. Your natural reaction is to push harder and harder on the brakes to try to make the car slow down. But of course nothing happens because the car speed is no longer controlled by the sliding tires. The normal feedback of car speed changing in reaction to your brake pressure is missing, so you apply more and more brake pressure to TRY to compensate until you run out of pedal. It is like that.

Are we talking about bias on the MOSFETs? Or are we talking about the voltage on them? Bias is a controlling voltage relationship that determines hom much they conduct together. That relationship remains whether the gates are sitting at about zero, or at about +25 or at about -18, or whatever. Whatever we have the outputs do, we maintain that bias relationship between the V+ and V- side gates. Now that is a different thing from the voltage on the gates. If we have a normal, working, properly biased amp, then the gate of the K135 will always be some amount of voltage more positive than the gate of the J50 - the voltage across P1. Now if we push the voltage to ground at P1 more positive, then the output bus follows and also goes more positive. Run P1 up to +20 and we ought to get about +20 on the speaker. But regardless of that, that space between the gatesd always remains the same.

If the amp is skewed, and wants to sit there with 20v on the output, it won;t usually hurt the amp unless you add a speaker load. With a load, now current will be drawn from that 20v. SO it won;t hurt to have the MOSFETs with +20v on their gates as long as no speaker load is present. That is not bias. What would kill them is to have +20 on the K135 gate and -20 on the J50 gate. That would be 40v across P1, not a couple volts. THAT would be wrong bias. So the voltage ACROSS P1 is your bias, not the voltage from it to ground.

Yess, of course you could take readings from each end to ground and do the arithmetic, but again, finding +20v or whatever on both ends of P1 is not the same as 20v across it.

[lowell]

Just replaced TR4 again, same result. -12.4v on P1. Yes I understand that the voltage across P1 is the bias, and that that voltage across P1 is different from measuring to ground. Ok, so, Ima put these MOSFETS back in and see. All I have here are 2SK1530 and 2SJ201.

[lowell]

Ooookaaaay. back to where we started. Have 2SK1530 and 2SJ201 in place. The voltage across P1 does change when it's adjusted. However, back to where we started w/ incorrect voltages on TR4/6 collectors.

P1 at 0ohms: TR4/6 C: -1.1v
P1 at 185ohms: TR4 C: -600mv
TR6 C: -1.1v

What if I were to add a diode between R22/D6 string? To turn on TR4 a bit more and shift the bias up?

[MarkusBass]

Steve may be right with the missing feedback loop. I used this trick in SVT4Pro, which has different topology. Here you have two differential pairs and the second pair is left without feedback and it is not referenced to the ground. But I did a simulation with a similar amp (Marshall) and I see that the output voltage is floating but it is not sticking to the rail. I get e.g. 1.9V and 3.2V (both positive) on the bias pot. And I can still change the voltage accros the pot (which would influence the bias with MOSFETs connected).
I'm still confused with the information you provide; previously you said that the voltage on P1 is -53V, now you say that it is still the same (but it's -12V this time). Which one is correct? And I'm not sure whether this voltage is in reference to the ground, or accross the pot.
You may solder back the MOSFETs but the pair that you have requires higher bias voltage to work correctly. But at least you will check whether you get positive and negative voltages on P1.
Sorry for the hint with resistors which may not work as expected with Trace.

Mark

[lowell]

Mark,
The bias is no longer stuck to the rail. And the -12v on P1 was when I reinstalled the 12v zener diodes, so P1 was still at the rail, but the zeners caused P1 to have -12v.

Anyway, I have the amp all wired per the schematic except for the subbed MOSFETS. We are now back where I started, but the collector voltages on TR4/6 not being positive and negative. For some reason this point in the biasing string is shifted down into the negative region. This is why I ask if I may try and turn on TR4 a bit more to shift it up - by inserting a series diode between its base and the positive rail.

[Enzo]

Let us look at your bias. Now with the whole circuit, the bias adjust works. YOu can adjust the gate spread from zero volts to 0.500v. SO zero to a half a volt is your bias range at the gates. Note that the bias has nothing to do with voltage to ground. If instead of -1.1 and -0.6 (0.500v), you had +23 and +22.5, you would STILL have a 0.500v bias.

What you are expecting is maybe +0.500 and -0.500v, centered around zero. But what you have is skewed negative 0.600v. That is not a bias problem, that is an offset problem. NO speaker load, right? Is there anay DC on the output? MAybe about a volt negative?

[lowell]

The output has -200mv of offset.

[Enzo]

Time for a load. Does putting a load on it knock that 200mv to zero? Or doesw it stay there?

[lowell]

with 8ohm resistive load the offset on output is still -200mv. However, the bias is now shifted down more to -1.7v and -1.2v

[lowell]

Enzo any thoughts on this?

[Enzo]

How does the amp operate? With load attached, so you have 200mv offset. How much current is drawn from the mains? Does the amp heatsink get HOT? Does it provide full power without clipping? Does it make clean undistorted output? SOmetimes things work OK even when small details are a bit skewed.

[lowell]

No it came back after the first repair just like this, and it sounded like it has crossover distortion. Shall I still check on those other things knowing this?

[lowell]

Well, thanks for your suggestion Enzo the thing sounds great and I don't see any crossover distortion on the output. Guess somewhere along the way it got fixed. This is a lesson in always "checking in" - seeing if there IS a problem. Good to know that an amp can work with offset. I didn't know that was the case.

[MarkusBass]

I've just measured almost identical power amp (2SK1530 and 2SJ201) and it has 30mV offset on the output. I think that the offset is caused by not matched transistors on the input of the amp (differential pair). Did you change them? Also the output transistors should be matched but I don't know whether this may cause the offset.

Mark

[lowell]

Markus I think I only replaced ONE of those diff pair transistors. That makes sense thank for the input!