Add Output Cap to Solid State Bass Amp
- Thread starter CJ
- Start date
Help Support GroupDIY Audio Forum:
Since it's not april I'll guess you are not joking about this, you should be in the mF range if you want a decent/good low end (I guess you do since it's a bass amp 8))
1000µF at 8Ω gives a -3dB point at 20Hz.
Then is matter to find a 1000µF cap (or bigger) which would be happy handling the current on the speaker, you probably want low ESR to avoid overheating it. This size of caps are for PS mostly, so you could find the ripple current rating, that number may say something. The bigger the value the lower the voltage across the cap and less it will care about higher currents.
The other thing, they will be polarized, so you should make the amp run always towards one end applying some DC at the input or use 2 caps to make them NP. I don't know how much abuse this caps can handle at AC when the inverted peaks hits them. The problem there is not the HF which won't present much voltage across the cap but the rumble at lower frequencies, which bass usually have some dirt in there. Here again, the bigger the cap the lower the cut frequency, lower the AC voltage across it, less you care about it.
JS
1000µF at 8Ω gives a -3dB point at 20Hz.
Then is matter to find a 1000µF cap (or bigger) which would be happy handling the current on the speaker, you probably want low ESR to avoid overheating it. This size of caps are for PS mostly, so you could find the ripple current rating, that number may say something. The bigger the value the lower the voltage across the cap and less it will care about higher currents.
The other thing, they will be polarized, so you should make the amp run always towards one end applying some DC at the input or use 2 caps to make them NP. I don't know how much abuse this caps can handle at AC when the inverted peaks hits them. The problem there is not the HF which won't present much voltage across the cap but the rumble at lower frequencies, which bass usually have some dirt in there. Here again, the bigger the cap the lower the cut frequency, lower the AC voltage across it, less you care about it.
JS
I couldn't tell if that's an 8 or a 2, I guess squared voltages are DC and circled are AC, in that case the cap is not even close to get negative voltage, very worst case, like shorted output ir very very low frequency, 26V RMS, ~37V peak, with 46V DC at the other end, I wouldn't go so far, but a would give it a few volts to be more conservative. If you tune it for 20Hz and you use that as your lower output freq you have 0.707 times the peak output voltage, you can't afford that in your setup as it will be missing a lot of headroom. If you go down to 5Hz you will have about 0.25 times the max peak at the cap, so giving it that bias will be enough, you will lose a bit of headroom but your cap will be much happier never having reverse bias in a "high" current application.
Not short circuit proof though but that would be harder, you should reference your speaker to the negative rail for example, but then you need a very clean negative rail, any ripple or injected noise will go directly to the speaker, this could be archived using that rail as reference instead of ground, without the need of having extra filtering at all, so the output will be following the negative rail (could be the positive using the cap the other way around) But then to be really short circuit proof you need the cap to be able to take the full voltage from rail to rail, so bigger rating cap. Maybe you prefer to blow a cap instead of the power transistors in a case of a short circuit so a smaller cap could be used, but they may fail as short circuit so you wouldn't save anything. Hard to get 150V 8000µF cap, also probably expensive. This would make the cap bullet proof, not the rest of the amp, it already have short circuit protection with Q3 and Q8. I'd sacrifice maybe 10V of headroom or even better if you can shift the PS rails 10V or 20V to one side so you don't lose any headroom and you have those 10V or 20V to bias the cap and protect it from reverse bias at AC peaks.
I hope not being putting c**p out there, correct me if I'm wrong. Tomorrow I have a test about FPGA and all that boring stuff, I shouldn't be playing with analog stuff now!
JS
Not short circuit proof though but that would be harder, you should reference your speaker to the negative rail for example, but then you need a very clean negative rail, any ripple or injected noise will go directly to the speaker, this could be archived using that rail as reference instead of ground, without the need of having extra filtering at all, so the output will be following the negative rail (could be the positive using the cap the other way around) But then to be really short circuit proof you need the cap to be able to take the full voltage from rail to rail, so bigger rating cap. Maybe you prefer to blow a cap instead of the power transistors in a case of a short circuit so a smaller cap could be used, but they may fail as short circuit so you wouldn't save anything. Hard to get 150V 8000µF cap, also probably expensive. This would make the cap bullet proof, not the rest of the amp, it already have short circuit protection with Q3 and Q8. I'd sacrifice maybe 10V of headroom or even better if you can shift the PS rails 10V or 20V to one side so you don't lose any headroom and you have those 10V or 20V to bias the cap and protect it from reverse bias at AC peaks.
I hope not being putting c**p out there, correct me if I'm wrong. Tomorrow I have a test about FPGA and all that boring stuff, I shouldn't be playing with analog stuff now!
JS
sahib
Well-known member
emrr said:
I just said "you probably want low ESR" because higher ESR would generate heat inside of the soup can, hard to get out... Not it wouldn't work but it would be more stressed and probably last for not so long. It wouldn't affect the performance of the amp much, I guess the normal ESR of a cap this size wouldn't degrade damping factor by much, I don't know.
JS
Why do the the fets go bad?
Do you match the MOSFETS before changing them?
I would build a circuit to do a multi-point match at different currents, requiring the MOSFETs to be mounted to a heatsink.
Voltage gate to source vs drain current
Do you match the MOSFETS before changing them?
I would build a circuit to do a multi-point match at different currents, requiring the MOSFETs to be mounted to a heatsink.
Voltage gate to source vs drain current
Putting a cap in series between an amp and speaker is more likely to protect the speaker than amp. I have mainly seen this done in amps to provide a bipolar AC output voltage from a single supply amp.
Vertical MOSFETs seem harder to use in linear power stages, and are unlikely to share well without matching (or additional circuitry to force sharing).
Good luck
JR.
Vertical MOSFETs seem harder to use in linear power stages, and are unlikely to share well without matching (or additional circuitry to force sharing).
Good luck
JR.
thanks for all the replies!
we had a bad zener that took out a transistor which shifted the load to one sdie of the amp,
replaced all the drivers and diodes and fets so we will fire it up,
this is for a customer so we can not buy 20 fets for each side and match, unfortunately,
would be best to get some 8 dollar Toshiba fets all matched and ready to go, people sell these on evilbay, if you are interested,
gonna add a cap also, these bass players crank the amp to 11, the speakers blow drags too much current and the fets pop,
there is some protection circuitry but i think it is for zero offset,
circled in red is the bad diode, these direct coupled amps have a domino effect, one thing pops and it starts a chain reaction of catastrophic and unwanted sequence of events,
hope the next amp that comes in is a twin reverb.
we had a bad zener that took out a transistor which shifted the load to one sdie of the amp,
replaced all the drivers and diodes and fets so we will fire it up,
this is for a customer so we can not buy 20 fets for each side and match, unfortunately,
would be best to get some 8 dollar Toshiba fets all matched and ready to go, people sell these on evilbay, if you are interested,
gonna add a cap also, these bass players crank the amp to 11, the speakers blow drags too much current and the fets pop,
there is some protection circuitry but i think it is for zero offset,
circled in red is the bad diode, these direct coupled amps have a domino effect, one thing pops and it starts a chain reaction of catastrophic and unwanted sequence of events,
hope the next amp that comes in is a twin reverb.
Attachments
The circled diode pair looks like a simple voltage clamp on the MOSFET gate drive. If that blew, the amp was being driven to play too loud... maybe rails are hotter than expected, or diodes were too small to survive clamping.
JR.
JR.
> here is an acoustic 470 with a cap
Your Ampeg runs on bi-polar +/- voltage. The output sits at zero V DC.
That 470 runs on *single* supply, +92V and zero, with output idling at +46V DC.
Your bi-polar plan really *must* use a bi-polar capacitor.
Ripple current is probably non-issue. Use a LOT of uFd, voltage rating better than supply rail (93V, 100V, 150V), and because you won't find a bipolar cap this big, use two back-to-back series.
Two 8,000uFd 150V will be two very large cans. Ample surface-area to throw-off whatever ESR heat the signal causes. But possibly hard to find that much space in a cool area and not real close to the input circuits.
Your Ampeg runs on bi-polar +/- voltage. The output sits at zero V DC.
That 470 runs on *single* supply, +92V and zero, with output idling at +46V DC.
Your bi-polar plan really *must* use a bi-polar capacitor.
Ripple current is probably non-issue. Use a LOT of uFd, voltage rating better than supply rail (93V, 100V, 150V), and because you won't find a bipolar cap this big, use two back-to-back series.
Two 8,000uFd 150V will be two very large cans. Ample surface-area to throw-off whatever ESR heat the signal causes. But possibly hard to find that much space in a cool area and not real close to the input circuits.
yeah, no, really spaced on the bi-polar thing thanks for the heads up!
gonna leave the caps out but clamp this thing down to 300 watts instead of whatever it is now, supposedly 350,
we have a resistive dummy load and 60 hz to inject, maybe back to back zeners with a pot in series as a varible clamp, find out where the ac signal sits at 300 watts and pick new zener values to make it happen without the pot,
gonna leave the caps out but clamp this thing down to 300 watts instead of whatever it is now, supposedly 350,
we have a resistive dummy load and 60 hz to inject, maybe back to back zeners with a pot in series as a varible clamp, find out where the ac signal sits at 300 watts and pick new zener values to make it happen without the pot,
ok this amp is back together and sounding great,
N channel IRFP240 fets matched up to within 0.1 milli-volts across the 0.47 ohm resistors right out of the tube,
P channel IRFP9240's were 5 mv apart, which is still good but no as exact as the N channel, this has been reported by other people matching these fets,
we replaced the 2N5415 and 3440 transistors with NOS Motorola cans off evilbay,
rail voltage were +/- 67 volts instead of 60, so we used taps on the toroid xfmr and managed to drop it back down to 60, this seems to be a common problem with offshore toroid pwr transformers,
since the prw xfmr is no longer using two primaries in parallel, we have to make sure we do not cook the winding by using a single pri and part of another pri in series,
so we say 360 watts with +/- 60 v-dc rails needs 360/120v-dc=3 amps,
use a 1:1 ratio and we see we need about 3 amps pri, should not be a problem,
so we save the fets 67-60 = 7 volts each rail = 14 volts at 3 amps = 42 watts,
this is the excess wattage generated by the out of spec pwr transformer,
and the zener clamps stay the same, so we limit the total power by 40 watts even if the guy cranks it up to compensate for lower rails,
so we hope the decreased power will add life to the fets and keep the amp from popping again during this lifetime,
N channel IRFP240 fets matched up to within 0.1 milli-volts across the 0.47 ohm resistors right out of the tube,
P channel IRFP9240's were 5 mv apart, which is still good but no as exact as the N channel, this has been reported by other people matching these fets,
we replaced the 2N5415 and 3440 transistors with NOS Motorola cans off evilbay,
rail voltage were +/- 67 volts instead of 60, so we used taps on the toroid xfmr and managed to drop it back down to 60, this seems to be a common problem with offshore toroid pwr transformers,
since the prw xfmr is no longer using two primaries in parallel, we have to make sure we do not cook the winding by using a single pri and part of another pri in series,
so we say 360 watts with +/- 60 v-dc rails needs 360/120v-dc=3 amps,
use a 1:1 ratio and we see we need about 3 amps pri, should not be a problem,
so we save the fets 67-60 = 7 volts each rail = 14 volts at 3 amps = 42 watts,
this is the excess wattage generated by the out of spec pwr transformer,
and the zener clamps stay the same, so we limit the total power by 40 watts even if the guy cranks it up to compensate for lower rails,
so we hope the decreased power will add life to the fets and keep the amp from popping again during this lifetime,
.025VDC / .47 ohm about = 53ma each device
As I posted before I would do a multpoint match
at least a two point match one at idle current, voltage gate source vs
and another match at a higher current
Curve tracing each output device might be a good thing to do
As I posted before I would do a multpoint match
at least a two point match one at idle current, voltage gate source vs
and another match at a higher current
Curve tracing each output device might be a good thing to do
