There is something very very wrong if removing R21 and 23 stop it working altogether. What exactly do you mean by that anyway ? Are your component designations as per the circuit diagram at the start. What DC volt drop have you across R26 and what DC drop across R27 with no load connected ?
We are trying to help.
We are trying to help.
Mooly said:
Oh I know you're trying to help and I am very grateful
[/QUOTE]
I solved that a few minutes ago; through all the fiddling around I had managed to work a connection loose, after restoring this the removal of R21 and 23 had interesting results.
Before when I reached the point of clipping the sound produced was REALLY severe, almost like the sound of a voice coil slamming into the back plate repeatedly. Now the clipping point is roughly the same, however it is far more benign in nature, it limits the cone excursion the same way as before and as the volume is increased beyond the onset of clipping, the distortion increases but its never what you could call unpleasant, like it used to be.
Yes.
R26 = 15.7mV
R27 = 14.8mV
Giving together 33.2mV, 33mV being the value Slone says to set the biasing at.
50mA bias through each FET is a bit low according to most designers.
Borbely suggests >=100mA per FET and >=500mA total bias for all parallel FETs.
My Sugden uses 75mA/FET and I think that's too low, but the sinks get too hot when I push it higher.
What are the voltage drops on the other two source resistors?
Are they 5% tolerance?
Confirm which FETs you built with.
Borbely suggests >=100mA per FET and >=500mA total bias for all parallel FETs.
My Sugden uses 75mA/FET and I think that's too low, but the sinks get too hot when I push it higher.
What are the voltage drops on the other two source resistors?
Are they 5% tolerance?
Confirm which FETs you built with.
The FETs I built with are the FETs I posted with the schematic, the EXICON 10N16 and 10P16.
Measuring all voltages again as the climate has changed slightly.
R26 = 15.8mV
R27 = 15.1mV
R28 = 16.9mV
R29 = 17.8mV
I can't say I know the tollerence of the resistors, they are wirewound high power jobbies. I could try and measure them if you want, but the small values could make it difficult to get an accurate reading.
Measuring all voltages again as the climate has changed slightly.
R26 = 15.8mV
R27 = 15.1mV
R28 = 16.9mV
R29 = 17.8mV
I can't say I know the tollerence of the resistors, they are wirewound high power jobbies. I could try and measure them if you want, but the small values could make it difficult to get an accurate reading.
What frequency are you testing at ?
As you don't have a scope how loud does it actually play with music, fairly loud, very loud etc. before distortion sets in.
I still don't like the look of the limiter, particularly around C14 and 15. It just does not look right to me. The drive at the MOSFET gates is "higher" than the signal at the output, the VGs value and this is non linear with load and frequency. Those caps are coupling this signal to the limiter transistors. I would take them out to try, or rather put them over the base emmiter junctions instead.
Edit- You say the original design uses 2SK1058 2SJ62. I think 250 watts into 4 ohm is pushing it a bit, even with a pair of these.
I know they are tough and immune to secondary breakdown, but normally 60 to 80 watts into 8 ohm would be a maximum for a single pair of these.
As you don't have a scope how loud does it actually play with music, fairly loud, very loud etc. before distortion sets in.
I still don't like the look of the limiter, particularly around C14 and 15. It just does not look right to me. The drive at the MOSFET gates is "higher" than the signal at the output, the VGs value and this is non linear with load and frequency. Those caps are coupling this signal to the limiter transistors. I would take them out to try, or rather put them over the base emmiter junctions instead.
Edit- You say the original design uses 2SK1058 2SJ62. I think 250 watts into 4 ohm is pushing it a bit, even with a pair of these.
I know they are tough and immune to secondary breakdown, but normally 60 to 80 watts into 8 ohm would be a maximum for a single pair of these.
Hi,
too late to measure now, but I have recommended matching the source resistors to allow accurate measurement of bias currents. This enables checking of the matching of your parallel sets of output devices.
100mA through a series string of 0r33 resistors allows you to pick out +-1% easily, by just measuring the voltage drops across each.
Then regroup in the selected sets and pull out the best matches and with care you can get around +-0.1% to +-0.2%.
You have a range of 17.9% and wirewounds are usually +-5%. That indicates that your FETs are not well matched at the quiescent current you have set up. But we are only guessing due to unmatched resistors.
too late to measure now, but I have recommended matching the source resistors to allow accurate measurement of bias currents. This enables checking of the matching of your parallel sets of output devices.
100mA through a series string of 0r33 resistors allows you to pick out +-1% easily, by just measuring the voltage drops across each.
Then regroup in the selected sets and pull out the best matches and with care you can get around +-0.1% to +-0.2%.
You have a range of 17.9% and wirewounds are usually +-5%. That indicates that your FETs are not well matched at the quiescent current you have set up. But we are only guessing due to unmatched resistors.
Here's some information on the Leach amp protection circuit, similar but different. Prof Leach may explain things better.
http://users.ece.gatech.edu/~mleach/lowtim/prot.html
regards
http://users.ece.gatech.edu/~mleach/lowtim/prot.html
regards
Leach explains it well.
But, his protection circuit omits the second slope, see the horizontal plot after the single slope.
This horizontal section allows the output device to operate outside it's SOAR.
The circuit needs one extra resistor in each half to create that second slope.
It is fitted in Jens' Leach clone.
Similarly, there is a missing capacitor in each half to allow short term transients to pass without triggering the protection prematurely.
But, his protection circuit omits the second slope, see the horizontal plot after the single slope.
This horizontal section allows the output device to operate outside it's SOAR.
The circuit needs one extra resistor in each half to create that second slope.
It is fitted in Jens' Leach clone.
Similarly, there is a missing capacitor in each half to allow short term transients to pass without triggering the protection prematurely.
Hi,
I think all this might apply bit more to BJT's and HEXFET'S rather than Lateral MOSFET's.
The Laterals are immune from secondary breakdown and the resistor you mention takes into account the supply rail voltage to prevent this problem. It's at high voltages - near but not exceeding VCE that these devices exhibit this problem- it's peculiar to BJT's.
Andrew's right about the transient limiting, where the caps are on the circuit is ( to my eyes ) tripping the protection prematurely, and with the effect of the diode as well at HF, it's half way to rectifying the drive and applying it to the base of the limiter.
Edit- Got that carried away forgot what I set out to say.
I would remove the protection completley and add gate source Zener and diode to each FET - this is the normal way of providing protection for FET's. I think it's 8.2 volt in series with a 1n4148 if you want to do this.
I think all this might apply bit more to BJT's and HEXFET'S rather than Lateral MOSFET's.
The Laterals are immune from secondary breakdown and the resistor you mention takes into account the supply rail voltage to prevent this problem. It's at high voltages - near but not exceeding VCE that these devices exhibit this problem- it's peculiar to BJT's.
Andrew's right about the transient limiting, where the caps are on the circuit is ( to my eyes ) tripping the protection prematurely, and with the effect of the diode as well at HF, it's half way to rectifying the drive and applying it to the base of the limiter.
Edit- Got that carried away forgot what I set out to say.
I would remove the protection completley and add gate source Zener and diode to each FET - this is the normal way of providing protection for FET's. I think it's 8.2 volt in series with a 1n4148 if you want to do this.
Hi,
When you say verticals are you referring to HEXFET's like the IRF types ? I believe there is a failure mechanism- that has been documented- that appears to show this type of FET does indeed seem to have a problem similar to secondary breakdown in BJT's.
The true lateral such as the 2SK/2SJ and the Exicon equivalents do not. The SOAR curve for these only has to take into account the device ratings and the thermal limits of the device.
When you say verticals are you referring to HEXFET's like the IRF types ? I believe there is a failure mechanism- that has been documented- that appears to show this type of FET does indeed seem to have a problem similar to secondary breakdown in BJT's.
The true lateral such as the 2SK/2SJ and the Exicon equivalents do not. The SOAR curve for these only has to take into account the device ratings and the thermal limits of the device.
Mooly said:
Currently a 55hz sine wave as that corresponds to a 4 ohm load into the paralleled XLS.
It depends on the music, as the XLS are used in an open baffle, there is a lot of Eq, so the amps are having to work reasonably hard in comparison to the rest of the system. It gets loud enough with some music and with others not so much. The issue is mainly how bass heavy, or how artificial the bass drum is. I've got several pieces of music where the bass drum is heavily processed, whenever the drum hits the protection trips in. I suppose the reason why I revisiting this is because it's acceptable but irritating the way it is.
With R21/23 removed, I first tested this with Music and thought the problem had been fixed, simply because the clipping, although still seeming to do the same job, was so benign it was hardly noticeable. The rest of the music, around the bass drum, was still increasing in volume, so it had the effect of seeming to be okay.
I'll give that a go and see what happens, are we talking with R21 and 23 in place or not?
Well I've got an adequate power supply and heat sinking to allow that I think, I wont get anywhere near Using 250 watts though in the current setup.
Hi,
I took the time to simulate this amp (according to your schematic) and it runs fine. Power output is easily 250 watts into 4 ohms before clipping, distortion less than .1% (could be better, I didn't take the time to find the correct models for the MOSFETS).
Protection circuitry works as designed too.
Are you sure you didn't make a mistake?
I took the time to simulate this amp (according to your schematic) and it runs fine. Power output is easily 250 watts into 4 ohms before clipping, distortion less than .1% (could be better, I didn't take the time to find the correct models for the MOSFETS).
Protection circuitry works as designed too.
Are you sure you didn't make a mistake?
MJL21193 said:
If I made a mistake it's only the mistake of circuit maker, for the schematic you're looking at is taken directly from the simulator. So if it works for you, unless you made a mistake that inadvertently corrected an error, it should work for me too. What simulation program are you using?
The one thing I have NOT done is compare Slones PCB to the schematic, I assumed it would be correct, but there is always the possibility...
I have checked and double checked the values of everything in the amplifers and can't see anything wrong, but again, there is always the chance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.