lepomis said:
I'll take your word for it, I did not want to register. I need to find my glasses I guessI wish I could help you more with your troubles, but I do not know enough about trouble shooting circuits to build that amplifier.
You need to register to view the image at full size? That's the last time I use them >.< When I load through the page, at the top of the image there is a yellow border, on the far left there is some text that reads "Original Image Size." You don't get that without registering?
Re post 12, IRF FET's will never work in this design. Two main problems, 1 Thermal Runaway. 2. They need a much higher Vgs-hence no bias current.
Offset voltages in the range you are getting suggest poor circuit design IMO rather than a fault in construction. Offsets of " up to quarter of a volt " were considered acceptable at one time, and to lower it the circuit had to be tweaked on an individual basis, usually by swapping transistors around.
Regards Karl
Offset voltages in the range you are getting suggest poor circuit design IMO rather than a fault in construction. Offsets of " up to quarter of a volt " were considered acceptable at one time, and to lower it the circuit had to be tweaked on an individual basis, usually by swapping transistors around.
Regards Karl
Yes I am unable to import the correct spice models into Circuitmaker so I chose something in the available parts. But now, to simplify matters, I have removed the output stage. I have built the design using the Exicon FETs.
I tried altering the value of the 470k resistor
(yes Andrew mentioned the time constant of the NFB and the input filter were miles apart) and tried 47k. Result? I've got less then 2mV offset in both amps. BUT they still clip prematurely. Now I have tried the adjustments Slone recommends in the book for altering the protection circuitry (increasing R24/25), but it has very little effect.
Do you think that the values in the protection circuitry are decent? Or is something wrong here?
I also tried deactivating the protection circuitry, rail per rail, both were activating prematurely this time.
I tried altering the value of the 470k resistor
(yes Andrew mentioned the time constant of the NFB and the input filter were miles apart) and tried 47k. Result? I've got less then 2mV offset in both amps. BUT they still clip prematurely. Now I have tried the adjustments Slone recommends in the book for altering the protection circuitry (increasing R24/25), but it has very little effect.
Do you think that the values in the protection circuitry are decent? Or is something wrong here?
I also tried deactivating the protection circuitry, rail per rail, both were activating prematurely this time.
the way the protection circuitry appears to work is the source current flowing out of Q13 (and its compliment) drops a voltage across R26. This voltage is then applied as the input to the voltage divider consisting of R24 and R21. When the output of this voltage divider (voltage dropped across R21) hits ~0.7V, it turns on Q9 which then sucks gate drive away from Q13, limiting its source current.
if my understanding of the circuit is correct, and with the values shown, this should happen at ~2.33A source current out of Q13. this trip point should be able to be adjusted by alteration of the voltage divider R24 + R21.
if my understanding of the circuit is correct, and with the values shown, this should happen at ~2.33A source current out of Q13. this trip point should be able to be adjusted by alteration of the voltage divider R24 + R21.
gain, that concurs with what Slone says in his description of this protection circuit. However he also adds, that because the bases of Q9 and Q10 are referenced to circuit common through two steering diodes D5 and D8, it allows the protection transistors to monitor the output current relative to the load impedance.
If we consider the 11 volts into 4ohms these can generate before showing signs of clipping, we see that this corresponds to 2.75 amps, which is close to your 2.33A calculation.
This also provides 18 volts into an 8 ohm load which corresponds with 2.25 amps before clipping.
Note that the load I'm using is a pair of XLS in parallel, im simply taking the impedance from Typhany's website and throwing the correct frequency through the speakers, then measuring it. So the impedance could be off by a small amount.
From an uneducated guess point it seems that the circuit isn't really detecting the load present at all and is just hard limiting at a relatively fixed value, that or something is completely wrong.
I have altered the value of R24/25 up to 200ohms to see if this has any noticeable effect and it really doesn't.
I also just measured the voltage drop across R21/23 driving 4ohms and the voltage reads around 5V before clipping starts.
If we consider the 11 volts into 4ohms these can generate before showing signs of clipping, we see that this corresponds to 2.75 amps, which is close to your 2.33A calculation.
This also provides 18 volts into an 8 ohm load which corresponds with 2.25 amps before clipping.
Note that the load I'm using is a pair of XLS in parallel, im simply taking the impedance from Typhany's website and throwing the correct frequency through the speakers, then measuring it. So the impedance could be off by a small amount.
From an uneducated guess point it seems that the circuit isn't really detecting the load present at all and is just hard limiting at a relatively fixed value, that or something is completely wrong.
I have altered the value of R24/25 up to 200ohms to see if this has any noticeable effect and it really doesn't.
I also just measured the voltage drop across R21/23 driving 4ohms and the voltage reads around 5V before clipping starts.
How are you testing and checking the operation of the limiter. Are you using a 'scope and a known load.
Analysis of the operation of the limiter is complicated by D5/8 and the associated resistors which form a divider to ground irrespective of the current flow in the 0.33 ohms. It is almost like a form of SOAR protection but the values look all wrong
What about this --- the action of R24 and R21 form a 10 to 1 divider that is ground referenced by D5. So as the output rises, irrespective of any load the limiter is going to cut in-- thought anyone ? 10 or 11 Volts peak to peak or so and thats the limit. That reference to ground makes it a voltage limiter to me.
Analysis of the operation of the limiter is complicated by D5/8 and the associated resistors which form a divider to ground irrespective of the current flow in the 0.33 ohms. It is almost like a form of SOAR protection but the values look all wrong
What about this --- the action of R24 and R21 form a 10 to 1 divider that is ground referenced by D5. So as the output rises, irrespective of any load the limiter is going to cut in-- thought anyone ? 10 or 11 Volts peak to peak or so and thats the limit. That reference to ground makes it a voltage limiter to me.
Mooly said:
I have explained how I measure the voltage with the assumed impedance, but the other point is a good one and may be rather important.
I do not yet have a scope so I'm not doing anything special wrt noticing the waveforms flattening, I assume at the peaks. All I do is listen to the sinusoidal wave form, the sound remains perfectly clear up until a point where boom it starts to sound like a saw tooth.
As I increase the volume past this point the buzzing gets worse as more of the waveform is distorted and the perceived volume gets larger, however, as expected, the loudspeaker displacement remains pretty much identical at least to the naked eye.
Slone says this circuit is almost 40 years old and has been used in many many amplifiers to great success, what specifically could be causing it to malfunction?
I assume the DC offset problem and the false activation of the protection circuit are completely independent?
Try removing R21 and R23. This will allow each FET to pass around 2.5amps max. See what difference that makes for a start.
If you need more output after that (and the heatsinks and PSU are rated correctly) you can then begin to think of adding 2 resistors too form a divider with R24/25.
Try removing those resistors first though.
If you need more output after that (and the heatsinks and PSU are rated correctly) you can then begin to think of adding 2 resistors too form a divider with R24/25.
Try removing those resistors first though.
hi 5th,
your thread has motivated me to drop by the library last nite on the way home from work and check out Mr. Slone's book, which i have been meaning to read for some time now. while i haven't made it through the whole book yet, i did find something that may be relevant on p 184, about halfway down the page, Mr. Slone states:
"[MOSFETs] are intrinsically diode protected against static discharge, immune to secondary breakdown, and are automatically current limiting. Ordinary fast blow fuses will actually protect lateral MOSFETs from overcurrent damage, in contrast BJT's will protect the fuse from damage every time."
if Slone is correct, and i have no reason to believe he's not, you could just scrap the protect circuity altogether and use appropriate voltage rail and/or output fuses.
your thread has motivated me to drop by the library last nite on the way home from work and check out Mr. Slone's book, which i have been meaning to read for some time now. while i haven't made it through the whole book yet, i did find something that may be relevant on p 184, about halfway down the page, Mr. Slone states:
"[MOSFETs] are intrinsically diode protected against static discharge, immune to secondary breakdown, and are automatically current limiting. Ordinary fast blow fuses will actually protect lateral MOSFETs from overcurrent damage, in contrast BJT's will protect the fuse from damage every time."
if Slone is correct, and i have no reason to believe he's not, you could just scrap the protect circuity altogether and use appropriate voltage rail and/or output fuses.
gain said:
I had thought of that myself, Slone has a very simple FET amp at the back of the book with no protection circuitry at all, the only "flaw" it has is that the rail fuses will blow on short circuit.
It would certainly fix the problem but I still find it very puzzling as Slone has actually built these amps. ( On page 134 Slone says "From a personal perspective , I think the 250 MOSFET design sounds great in both domestic and public address applications.") Now I am willing to accept a schematic error for the input filter, using a 47k instead of a 470k has fixed the DC offset problem. But the output protection failing also? Something just doesn't add up.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.