Amplifier Troubleshooting.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've been here before and with the very same amplifier, Slones design 11.8 in the high power audio amplifier construction manual, although I have never actually had them working completely as intended.

The problems can be summed up by two things.

1) High DC offset.
2) Protection circuitry tripping way before it should.

I have contacted Slone on a few occasions trying to solve this, but he failed to respond when I replied with the results of his tests.

First he asked me to make sure all the zenners were of the correct wattage, which they were.

He then asked me to check the voltage across the zenners and also the voltage across R16 and R18. Everything checked out normal, to which I responded, but I never got a reply, and even after replying for second time a while later, still no response.

I have also tried simulating the design in Circuit Maker 2000, but ran into a brick wall when trying to import individual spice models, I cannot figure out how to do this in a simple fashion, the program seems to be able to handle .lib files, but not .mod. Nevertheless I tried using the FETs used in the passlab amps and other transistors for pre drivers, but only got mixed results.

The DC offset and the biasing seem to be somewhat linked, I do not know if they should be, but in my amps they are. If I alter the value of P1 to one extreme the DC offset reaches about -200mV(this is as low as it goes), but I cannot get anything above about 5mV bias across the output FETs. Swinging the value of P1 around a bit, increasing the DC offset to around -280mV, allows the biasing to function where I can get the specified 33mV.

A while ago I tried disconnecting the protection circuitry, one rail at a time and from memory the side that's causing the over protection is the negative one, coincidence with the -ve DC offset? I don't know.

Slone says these amps are capable of delivering in excess of 250 watts into a 4ohm load, which my pair don't even come close to. Driving a pair of Peerless XLS at 55hz (where peerless says each are at 8ohms) the amps protection decides to start working at around 11 Volts RMS.

The XLS are run Open baffle and obviously the amps clip way before I run out of xmax when playing normal music. If I disable the protection circuitry I can hit Xmax easily, removing the protection circuitry isn't the answer.

Aside from the high DC offset and limited output power the amps sound just fine and I do use them, however I have always used them with the notion of getting them properly working one day.

Having talked about this a while ago on DIYaudio, most of which discussed had little effect and with getting nowhere with Slone, I've almost given up expecting them to function. I have heard of Slone posting a couple of faulty designs in the past, but this amp he's actually built, there's a picture of the damned thing in the book, so it obviously functioned for him. (Why build something and never test it?)

I would obviously prefer one final attempt at getting these working, over buying/making another design, but I almost get the impression that something really isn't right.

So here I am asking one final time, to see if any of you electrical whizzes can come up with anything.

Many thanks,

Matt.

P.S. - If I cannot get these to work, what other designs are there that would work well with a 60V power supply? Preferably with the output devices mounted on either side of the PCB, that way I could just recycle the cases and PSU.
 
Hi 5th,

If you can, please post the schematic of the amp you built, will be a big help.

If the circuit has a differential amp for its input stage (i believe most of Mr. Slone's designs do) then the dc offset may be due to a current imbalance in the differential stage (LTP). try beta-matching the LTP transistors, especially the ones for the current mirror, if applicable. you could also try replacing the emitter degeneration resistors in the LTP with a trimmer. say you have two 100R for emitter resistors, replace with a 200R trimpot with the wiper connected to the 'tail' (toward the constant current source). set the trimpot in the middle (100R on each side of the wiper) then power up with a voltmeter tied across the output, then use the trimpot to 'tune' away any dc offset.

at any rate post schematic and i'm sure one of the guys here will have it figured out in 2 seconds.
 
Alright here's the schematic, bare in mind accuracy is dependant on my accuracy in copying it into circuit maker;) I've checked everything thought and it appears to be correct.

In copying the schematic into paint to create a gif, some of the black blobs showing, what connects to what, were removed, I painted them back in, but there is of course the chance I missed one or two.

The Zenners at D1 and D2 are 18 Volt 1/2 watts. D3 and D4 are 3.9v 1/3 watt.

The first trimpot (500ohm 10% R13) is P1. The second trimpot (500ohm 60% R20) is P2 in Slones schematic and is quite obviously the one used for setting the bias. The % simply refers to how far around the trimmers are set, I assume 50% being midway.

http://www.turboimagehost.com/p/325123/amp.GIF.html
 
Hi 5th,
allowing 2mA through Q8 gives 300mV across R19 and it's complement.
Allowing 600mVbe across Q8, gives a total voltage of 900mV across R11.
Allowing 1mA through each half of the long tail pair requires 910r for R11 and it's complement.
R4 & R5 have 18V and pass 2mA therefore to match 910r they need to be 9k1.

Time constant of NFB loop is 120mS (1k2 & 100uF).
Time constant of Input Filter is 2350mS (470k & 5uF).
This reversal and mismatch could lead to a possible LF instabilty.
50k & 2u2F at the input is a possible alternative, RC=110mS.
If the source also has a 10uF at it's output then RC drops to 90mS.
Now set the NFB>=1.4*110mS>=150mS.
R=1k2, C>=130uF, use 150uF.

Note, these example values are not fixed, but the method shows where you have choices and how your selection cascades through the various stages of the design.
 
Administrator
Joined 2007
Paid Member
Hi,
It all looks a bit suspect to me. The protection circuit, the values dont look right at first glance. 250Watt/4ohm that's serious power, one quarter killowatt, thats 32 volts RMS or 45 volts peak into 4 ohms. A base - emmiter junction over 0.33 ohms, admittedly modified by the effect of R21/22/23 and D5 and D8, what's that going to limit at, just over 2 amps + the same for the other unmonitored output FET. Not enough ?

Edit- I have nearly deleted this post, as I am not sure without sitting down and working it through. Anyone care to do the sums.
Mad thought, are your 0.33 ohm resistors suitably rated as some types can dramatically increase in value as they overheat without failing, and return to normal when cold. This would cause the limiter to cut in early.
 
The main problem I have at the moment is importing spice .mod files into CircuitMaker 2000. I've got the files both for the 669/649s and the output FETs, I just don't have any idea how to import them.

I would obviously like to get the simulation working perfectly first, but at the moment everything seems rather screwy.

With the default values in Slones diagram I get 1.4 volts DC offset with both trimmers at 50%.

Leaving all other resistors as is and changing the NFB cap + input filter lowers this dramatically to an almost negligible 50mV.

I am also getting lower distortion, more accurately reproduced sine waves at 20khz etc.

I still however cannot change the biasing in the simulation, which obviously means I cannot make assumptions based on the overall distortion performance, but altering the input filter seems to have made a huge difference.

Altering the other resistor values prescribed just screws everything up, the 20khz sine wave goes to pot and the DC offset shoots up to 536mV.

This all could be being ruined by my quick choice of FETs and the replacement for the 669/649s.

I chose IRFP240 and IRFP9240 for the FETs and ECG186/7 for the 669/649. I'm flying blind on those :xeye:

I'm currently tempted to simply change the input filter/NFB thing and see if that makes any appreciable difference.
 
Mooly said:

Edit- I have nearly deleted this post, as I am not sure without sitting down and working it through. Anyone care to do the sums.
Mad thought, are your 0.33 ohm resistors suitably rated as some types can dramatically increase in value as they overheat without failing, and return to normal when cold. This would cause the limiter to cut in early.

Yes they are 7 watt.
 
AndrewT said:
before you ask the simulator the wrong questions and thus get the wrong answers, try understanding the circuit and the way it's supposed to work using manual calculation.

I don't wish to be rude, but I don't have the energy to learn how this whole thing works, I have chronic fatigue and a plethora of other things that need doing prior to us moving house in a few weeks time. I have always intended on learning some of the fundamentals but due to my health decline a few years ago there have always been other things that I'd rather spend my limited energy on, such as loudspeakers.

At this time I do not want to understand how they work, unless someone could just implant the data into my brain, I just want them to work, which I am sure you can understand. I am picking up bits and pieces from this discussion already, but I asked here because I know there are people far more knowledgeable then myself and I wished for their assistance in fixing the amplifier, I don't see anything wrong with this.
 
Slone has the the input stage for this amplifier described in the same book and just to simplify things I've removed the output stage, which I assume isn't particularly at fault, even though I'm sure its far from being perfect.

Now.

Bottom line is this I don't have the Hitachi 669/649 which *could* create an issue.

Slones input stage + VA does NOT work with the schematic provided, it just fails in the same way the amplifier does. Changing the values of the NFB and the input filter change all this and it starts to operate, from what I can tell, perfectly.

P1 can set the DC offset to Zero if adjusted perfectly, adjust one way it increases, adjust the other way it decreases, keep going and we end up getting -ve DC offset.

Running distortion analysis on this gets figures within the ballpark of what Slone does and the overall amplifier gain is correct too.

Working with what I understand currently this seems to be working.
 
lepomis said:
I noticed that one of the resistors in the protection circuit is different than what slone has in his book, but you may have changed it for some reason. Just something I noticed, but not sure if it would cause your trouble.

Would you mind elaborating? All the resistors in the schematic I posted and the one in Slone's book all match up. The design I'm talking about is on page 343.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.