Well,
the amp sitting on the bench with no source or speakers is 100% stable. Haven't tried it with source but not connected to speakers. Will try that tonight too, although it won't be long till I run out of source resistors.
Gonna order more parts from mouser tonight.
more later.....
Fran
the amp sitting on the bench with no source or speakers is 100% stable. Haven't tried it with source but not connected to speakers. Will try that tonight too, although it won't be long till I run out of source resistors.
Gonna order more parts from mouser tonight.
more later.....
Fran
woodturner-fran said:
Hmm, based on my limited under standing of the physical world, I think when driving a speaker, a higher amount of heat in AB may be generated to the amp. In my business, faulty connections fail when more stuff passes through it. So I assume that the speaker presents a different situation to the amp than running with no load.
I would think something as simple as a bad solder joint could be at play, or some other resistor failing causing a higher load to the amp. It works until limits are pushed.
two further tests before you risk connecting your speakers.
test 1.) connect a music source and play quiet music with the output still open circuit. How long does it play before blowing the resistors?
test 2.) connect an 8r0 load to the speaker terminals. connect a source and play quiet music into the resistor load. How long till the resistors fail?
During your tests over several hours I suggest you measure the output offset and check the average AC voltage at the output (~0.5Vac to 2Vac).
test 1.) connect a music source and play quiet music with the output still open circuit. How long does it play before blowing the resistors?
test 2.) connect an 8r0 load to the speaker terminals. connect a source and play quiet music into the resistor load. How long till the resistors fail?
During your tests over several hours I suggest you measure the output offset and check the average AC voltage at the output (~0.5Vac to 2Vac).
Oscillations
CViller, thank you again for providing us diyers these nice boards.
I'm not sure what you mean by dangling, perhaps because the resistor lead goes past the end of the board and is bent around it?
That is just temporary to keep it from flopping around. When the mosFets are in place and soldered on, the resistor lead will be soldered right at the mosFet giving only a couple of millimeters of lead.
I've read the suggestion in another thread on mosfet amps oscillating,
that splitting the resistance in half and putting one at the driver transistor on the board and the other at the mosfet. This was the suggested by a person that worked for a living designing (if I remember correctly) rf-equipement. That would be required when there is a wire from the pcb to a mounted mosfet.
...just my
--From the Pass F5 Service Manual...............
A caveat is in order here – this is a very wide band amplifier with a high input impedance. In order to prevent the output
voltage from bleeding back to the input at very high frequencies (thus making a fi ne power oscillator), keep the input and
output cables separate, and don’t externally connect the speaker ground to the input ground. Good ground shielding on the
input cables is important, and caution is called for in using Litz and other specially low inductance / high capacitance cables.
I have not seen a specifi c example of a problem, but historically it is to be expected when an amplifi er’s bandwidth exceeds
200 KHz. If the amp makes funny noises, runs extra hot, or blows fuses, this might be an indicator of such an issue.
cviller said:
CViller, thank you again for providing us diyers these nice boards.
I'm not sure what you mean by dangling, perhaps because the resistor lead goes past the end of the board and is bent around it?
That is just temporary to keep it from flopping around. When the mosFets are in place and soldered on, the resistor lead will be soldered right at the mosFet giving only a couple of millimeters of lead.
I've read the suggestion in another thread on mosfet amps oscillating,
that splitting the resistance in half and putting one at the driver transistor on the board and the other at the mosfet. This was the suggested by a person that worked for a living designing (if I remember correctly) rf-equipement. That would be required when there is a wire from the pcb to a mounted mosfet.
Zen Mod said:
...just my
--From the Pass F5 Service Manual...............A caveat is in order here – this is a very wide band amplifier with a high input impedance. In order to prevent the output
voltage from bleeding back to the input at very high frequencies (thus making a fi ne power oscillator), keep the input and
output cables separate, and don’t externally connect the speaker ground to the input ground. Good ground shielding on the
input cables is important, and caution is called for in using Litz and other specially low inductance / high capacitance cables.
I have not seen a specifi c example of a problem, but historically it is to be expected when an amplifi er’s bandwidth exceeds
200 KHz. If the amp makes funny noises, runs extra hot, or blows fuses, this might be an indicator of such an issue.
Hi all,
First to answer a few of the questions/comments:
1. Cviller: Gate stoppers are right on the gates short resistor leads just 2mm above the wide part of the mosfets leg.
2. Ichiban: ICs and speaker cables are kept well separated. Internally in the amp speaker grounds, PS ground are brought together at the speaker terminals. Input grounds go to the boards. circuit ground tied to chassis earth through back to back diodes paralleled with 0.01uF cap and 15R resistor. Summary: i think I have wired the amp correctly... but am open to correction!
3. AndrewT: partially answering your Q - I have monitored DC offset and it stays low throughout until the resistors go. Bias is a bit jumpy as music plays, might flick up and down by maybe 30mV, really quickly. Settles within a second or 2 of no music playing.
So jobs to do tonight:
1. Babowana: I will measure Vgs of the jfets
2. Fit new resistors and get the amp running again.
3. ZenmodLook at output on CRO, although I'm not sure what exactly to look for!
4. AndrewT Test for time amp is viable with no speaker connected and with resistor in place.
Lastly, I am very tempted to fit the fairchild mosfets and see if that sorts out this issue, although I would like to find out whats wrong.
If anyone has any other ideas, feel free to post them up here. I will also take some pics tonight and post them
Thank you all for your help and support!
Fran
First to answer a few of the questions/comments:
1. Cviller: Gate stoppers are right on the gates short resistor leads just 2mm above the wide part of the mosfets leg.
2. Ichiban: ICs and speaker cables are kept well separated. Internally in the amp speaker grounds, PS ground are brought together at the speaker terminals. Input grounds go to the boards. circuit ground tied to chassis earth through back to back diodes paralleled with 0.01uF cap and 15R resistor. Summary: i think I have wired the amp correctly... but am open to correction!
3. AndrewT: partially answering your Q - I have monitored DC offset and it stays low throughout until the resistors go. Bias is a bit jumpy as music plays, might flick up and down by maybe 30mV, really quickly. Settles within a second or 2 of no music playing.
So jobs to do tonight:
1. Babowana: I will measure Vgs of the jfets
2. Fit new resistors and get the amp running again.
3. ZenmodLook at output on CRO, although I'm not sure what exactly to look for!
4. AndrewT Test for time amp is viable with no speaker connected and with resistor in place.
Lastly, I am very tempted to fit the fairchild mosfets and see if that sorts out this issue, although I would like to find out whats wrong.
If anyone has any other ideas, feel free to post them up here. I will also take some pics tonight and post them
Thank you all for your help and support!
Fran
Re: Oscillations
A question on this then,
For the F4, I grounded the signal input ground and speaker output together on my jumper between the power supply output end.
For this amp, I should do otherwise?
Right now, with F5 -with my little FR speakers, with the signal, speaker and chassis ground all together in a clip lead composite, I have no problems. Nelson mentions externally grounded....So I don't think there is a problem if the grounds are stared together in the amp.
Am I wrong?
Tea-Bag.
ichiban said:
...just my
A caveat is in order here – this is a very wide band amplifier with a high input impedance. In order to prevent the output
voltage from bleeding back to the input at very high frequencies (thus making a fi ne power oscillator), keep the input and
output cables separate, and don’t externally connect the speaker ground to the input ground. Good ground shielding on the
input cables is important, and caution is called for in using Litz and other specially low inductance / high capacitance cables.
I have not seen a specifi c example of a problem, but historically it is to be expected when an amplifi er’s bandwidth exceeds
200 KHz. If the amp makes funny noises, runs extra hot, or blows fuses, this might be an indicator of such an issue.
A question on this then,
For the F4, I grounded the signal input ground and speaker output together on my jumper between the power supply output end.
For this amp, I should do otherwise?
Right now, with F5 -with my little FR speakers, with the signal, speaker and chassis ground all together in a clip lead composite, I have no problems. Nelson mentions externally grounded....So I don't think there is a problem if the grounds are stared together in the amp.
Am I wrong?
Tea-Bag.
Re: Oscillations
That's good, but I seriously doubt you'll get any benefit from mounting the gatestoppers like this. With normal placement you only have a centimeter or so to the resistor, which only becomes interesting way up in the MHz range. Good cabling on the input/output is more important with this amp.
ichiban said:
CViller, thank you again for providing us diyers these nice boards.
I'm not sure what you mean by dangling, perhaps because the resistor lead goes past the end of the board and is bent around it?
That is just temporary to keep it from flopping around. When the mosFets are in place and soldered on, the resistor lead will be soldered right at the mosFet giving only a couple of millimeters of lead.
I've read the suggestion in another thread on mosfet amps oscillating,
that splitting the resistance in half and putting one at the driver transistor on the board and the other at the mosfet. This was the suggested by a person that worked for a living designing (if I remember correctly) rf-equipement. That would be required when there is a wire from the pcb to a mounted mosfet.
That's good, but I seriously doubt you'll get any benefit from mounting the gatestoppers like this. With normal placement you only have a centimeter or so to the resistor, which only becomes interesting way up in the MHz range. Good cabling on the input/output is more important with this amp.
Babowana Vgs of Q1 (sk170) is 20.86V with R11/12 blown and trimmers left as they were. Vgs of Q2 2SJ74 is 20.68V
I fitted new R11 and R12 and one thing I did do thats maybe just a stab in the dark....
in my setup the mosfets are on short leads (30mm) with the stopper right on the pin. But I also populated the stopper on the PCB too, but had shorted it with some wire. So i cut that wire, so effectively now I have 2 x 47R in series for stoppers, one on the PCB and one on the gate.
So far amp is warming up but bias is behaving normally so far.
More later....
Fran
I fitted new R11 and R12 and one thing I did do thats maybe just a stab in the dark....
in my setup the mosfets are on short leads (30mm) with the stopper right on the pin. But I also populated the stopper on the PCB too, but had shorted it with some wire. So i cut that wire, so effectively now I have 2 x 47R in series for stoppers, one on the PCB and one on the gate.
So far amp is warming up but bias is behaving normally so far.
More later....
Fran
Woodturner-fran
Sounds like you have the wiring sorted out very well.
I was just throwing out a hint mainly reguarding the
possible cable inductances/capacitances. I'm not sure
about the electrostats.
Tea-Bag
I'll have to bow to those more knowledgeable on the grounding.
CViller
"With normal placement you only have a centimeter or so to the resistor, which only becomes interesting way up in the MHz range."
Just being careful. When I built the Borbely mosfet amps I recall warnings of oscillations in the
MHz range. Audio Amature had a number of fixes for that. It wasn,t until recently that I read
about using the stoppers at the drivers and outputs to mitigate the chance of oscillations due to
the inches of wire from board to output mosfets (inductance is the enemy).
Yes, the F5 boards make for a very compact layout helping enormously.
Sounds like you have the wiring sorted out very well.
I was just throwing out a hint mainly reguarding the
possible cable inductances/capacitances. I'm not sure
about the electrostats.
Tea-Bag
I'll have to bow to those more knowledgeable on the grounding.
CViller
"With normal placement you only have a centimeter or so to the resistor, which only becomes interesting way up in the MHz range."
Just being careful. When I built the Borbely mosfet amps I recall warnings of oscillations in the
MHz range. Audio Amature had a number of fixes for that. It wasn,t until recently that I read
about using the stoppers at the drivers and outputs to mitigate the chance of oscillations due to
the inches of wire from board to output mosfets (inductance is the enemy).
Yes, the F5 boards make for a very compact layout helping enormously.
Ok... an update.
1. I connected up a different CDP and set it playing. Almost immediately the bias jumped all over the place and the resistors started to cook big time.
2. That was with no speakers connected but using the full line output of the CDP. Last night it would have being playing with a much lower input level.
3. So definitely somehow this thing is unstable, despite having 2 gate stoppers and also it wasn't even at full bias voltage, or warmed up fully. However the level of the signal input would have been much higher tonight (direct from CDP) compared to last night (rather low level through preamp).
4. Temp of the mosfets was about 70 degC just before this (taken on the metal tab). When fully warmed up last night this was 80deg C.
5. So something here is sending bias wild (ran up and down erratically between 0.5-1.5V when resting bias was 0.35V), causing huge current draw and burning out the resistors.
So to answer more Qs:
AndrewT: only took music playing for a few seconds to see this... last night with speakers connected it took much longer.
Zenmod: I can't really hook up the CRO when this is happening, i wouldn't be fast enough.
What you guys think??!? Other channel gets the same signal and is absolutely stable.
Fran
1. I connected up a different CDP and set it playing. Almost immediately the bias jumped all over the place and the resistors started to cook big time.
2. That was with no speakers connected but using the full line output of the CDP. Last night it would have being playing with a much lower input level.
3. So definitely somehow this thing is unstable, despite having 2 gate stoppers and also it wasn't even at full bias voltage, or warmed up fully. However the level of the signal input would have been much higher tonight (direct from CDP) compared to last night (rather low level through preamp).
4. Temp of the mosfets was about 70 degC just before this (taken on the metal tab). When fully warmed up last night this was 80deg C.
5. So something here is sending bias wild (ran up and down erratically between 0.5-1.5V when resting bias was 0.35V), causing huge current draw and burning out the resistors.
So to answer more Qs:
AndrewT: only took music playing for a few seconds to see this... last night with speakers connected it took much longer.
Zenmod: I can't really hook up the CRO when this is happening, i wouldn't be fast enough.
What you guys think??!? Other channel gets the same signal and is absolutely stable.
Fran
There's something puzzling me --
Chart 7 and Chart 11 in the audioXpress article seem to be a little anamolous -- if we look at Chart 7, the THD% @ 2W into 8 ohms is 0.0027% -- but in Chart 11, the same data point has a THD% of 0.006% Expliquez cela a mois, s'il vous plais.
FWIW, in an F5 I just put together this afternoon, the 1kHz, 2W, THD% was 0.006%. I know we're counting the number of angels which can fit on the head of a pin, Aquinas, but.....
Chart 7 and Chart 11 in the audioXpress article seem to be a little anamolous -- if we look at Chart 7, the THD% @ 2W into 8 ohms is 0.0027% -- but in Chart 11, the same data point has a THD% of 0.006% Expliquez cela a mois, s'il vous plais.
FWIW, in an F5 I just put together this afternoon, the 1kHz, 2W, THD% was 0.006%. I know we're counting the number of angels which can fit on the head of a pin, Aquinas, but.....
Hi, Fran
It seems to be beyond my ability.
I understand from you that the amp is ok with dc bias, but it is problem with music signal at the input. Is it parasitic oscillation? If so, I would try 220 ohms of the mosfet gate resistors or greater values, replacing the original 47 ohms.
Is there any possibility of positive feedback?
If I am sure of all others, I would try to replace jfets and mosfets as bobodioulasso says.
Good luck!
It seems to be beyond my ability.
I understand from you that the amp is ok with dc bias, but it is problem with music signal at the input. Is it parasitic oscillation? If so, I would try 220 ohms of the mosfet gate resistors or greater values, replacing the original 47 ohms.
Is there any possibility of positive feedback?
If I am sure of all others, I would try to replace jfets and mosfets as bobodioulasso says.
Good luck!
