Hi All
So a friend accidently blew this amp up due to a short in his speaker cabling, Upon inspection it uses BUZ900P and BUZ905P outputs, this amp had been worked on before because the transistors had been changed before and they did not look original.
Finding a schematic for this amp is near impossible, although i managed to find a shcematic for the A2 which is similar.
Anyway i gave the amp a good going over, all new electrolytics, new emmiters (as one had been taken out). I luckily had some spare buz900ps to test the amp.
Now with the original 900P/905P It worked with no issues, i could adjust the idle current/bias to 60mv (across the emmiters) and above i had it playing for hours. My friend decided he wanted all new outputs as the original ones were a mismatch and had been replaced in the past.
I decided based on research to use Exicon ECX10P20R/10N20R outputs. I used a device to measure them and match them best i could.
To make sure they were fine i did one channel at a time, here is my issue. With one channel running 10P20R/10N20R and the other with BUZ900/905 it worked without issues.
As soon as i replaced all the outputs on both channels to exicons i get a low level hum from the speakers and output if i try and adjust the idle current over 40mv?
The only way to stop the hum is to have it running at say 35-40mv, if i got higher i get an annoying hum from the speakers. At first i thought i had damaged something so checking voltages showed nothing bad.
I replaced the outputs back to the buz900/905s and no issues or hum.
So my question is what is going on? i have a few ideas.
1. Could have an issue with the new mosfets? maybe one or two are faulty
2. They are not matched for the amp and not a good replacement choice?
3. Perhaps the gate voltages are different and the driver transistors need to be updated to match the new outputs?
I could leave it at 40mv but then its not running at class A and i really want to get this running as it should. Any advice or tips would be appreciated.
Thanks
So a friend accidently blew this amp up due to a short in his speaker cabling, Upon inspection it uses BUZ900P and BUZ905P outputs, this amp had been worked on before because the transistors had been changed before and they did not look original.
Finding a schematic for this amp is near impossible, although i managed to find a shcematic for the A2 which is similar.
Anyway i gave the amp a good going over, all new electrolytics, new emmiters (as one had been taken out). I luckily had some spare buz900ps to test the amp.
Now with the original 900P/905P It worked with no issues, i could adjust the idle current/bias to 60mv (across the emmiters) and above i had it playing for hours. My friend decided he wanted all new outputs as the original ones were a mismatch and had been replaced in the past.
I decided based on research to use Exicon ECX10P20R/10N20R outputs. I used a device to measure them and match them best i could.
To make sure they were fine i did one channel at a time, here is my issue. With one channel running 10P20R/10N20R and the other with BUZ900/905 it worked without issues.
As soon as i replaced all the outputs on both channels to exicons i get a low level hum from the speakers and output if i try and adjust the idle current over 40mv?
The only way to stop the hum is to have it running at say 35-40mv, if i got higher i get an annoying hum from the speakers. At first i thought i had damaged something so checking voltages showed nothing bad.
I replaced the outputs back to the buz900/905s and no issues or hum.
So my question is what is going on? i have a few ideas.
1. Could have an issue with the new mosfets? maybe one or two are faulty
2. They are not matched for the amp and not a good replacement choice?
3. Perhaps the gate voltages are different and the driver transistors need to be updated to match the new outputs?
I could leave it at 40mv but then its not running at class A and i really want to get this running as it should. Any advice or tips would be appreciated.
Thanks
Attachments
The Exicon MOSFETS can work well with the right parts connected to gate, and aluminum oxide insulators to decouple capacitance from the heat sink. There is a thread on this here, search for "Revisiting lateral MOSFET stability"
Probably RF oscillation(*).
It important when bringing up a new/modified amplifier to limit (or at least monitor) the current being drawn from the rails. The onset of oscillation would here show up as a jump-step in current consumption as you vary the bias.
One thing I note is there are source-resistors in that circuit - laterals are not normally used with source resistors, and if they are they must be non-inductive (not wire-wound) - inductance in the source lead of a MOSFET is a bad thing for stability.
The BUZ90x and ECX10P/N20 are close matches, but not identical, which is probably what's tipping this amp over the edge - its probably only barely stable with the original BUZ90x's anyway. Perhaps a higher stopper resistor on the gate will help? Check the rail caps aren't dried out.
(*) When a MOSFET output stage oscillates it is at very high frequencies (MHz), at which it behaves highly non-linearly, and the PSRR drops to nothing, allowing rail ripple straight through into the output - which is I think what you are hearing as hum. Do you have a 'scope?
I have a stock A220 with no modifications I bought as new in about 1998. I'll put a small pair of speakers on and a scope later to show you what the original A220 gives. I'll be able to also give @liu735 the voltage from the toroidal. I need to investigate the issue that it seems to have with the slow start. A couple of years ago I tried asking MF for the schematics previously but they sold the company east and so there's no expertise/documentation available.
If the thing dies I'll rip out the internals and upgrade the thing. I liked it's sound but it's missing some clarity now days compared to some of the kit.
If the thing dies I'll rip out the internals and upgrade the thing. I liked it's sound but it's missing some clarity now days compared to some of the kit.
Last edited:
Well I had the scope attached across the transformer with AC coupling - this showed 100V peak-to-peak (given the main caps are 50V that makes sense.
I have the amp connected to a small pair of 4 ohm CHN-50s. The amp is advertised to be able to drive 4 ohm (previously it's only been connected to 8ohm castle harlechs). On powering up a second time to check the current draw across the slow start - I was greeted with a light show and one of the 250V T6.5A mains power fuses has vaporised. No scope probe ground leads connected. Checking the resistance between the probes just attached to the scope show a 20.3Mohm connection (the multimeter has a 200M setting) so the scope probes didn't cause the short..
I had channel 3 & 4 connected over R213 - DC coupled, and the scope showed a 50V with 25V on other side. R213 seems to be a 50ohm resistor, so 25/50 = 0.5Amps through it giving 7.5W but it's not switched out. Measuring R213 in circuit shows 342ohms.. So something isn't right and I think I need to take the circuit board out and have a look.
Removing both fuses shows the transformer is fine still, it's delivering 104V peak-to-peak unloaded on both sides of the secondary winding.
So something is seriously up with the amp.. None of the FETs show outwards signs such as smoke, crack or discolouring (even of the dust). Not surprised after 24 years with the cooking high temps the amp runs at (hotter than you can touch on the top) that something has shorted.. well at least it gives me something to investigate/fix/rip&upgrade.
I have the amp connected to a small pair of 4 ohm CHN-50s. The amp is advertised to be able to drive 4 ohm (previously it's only been connected to 8ohm castle harlechs). On powering up a second time to check the current draw across the slow start - I was greeted with a light show and one of the 250V T6.5A mains power fuses has vaporised. No scope probe ground leads connected. Checking the resistance between the probes just attached to the scope show a 20.3Mohm connection (the multimeter has a 200M setting) so the scope probes didn't cause the short..
I had channel 3 & 4 connected over R213 - DC coupled, and the scope showed a 50V with 25V on other side. R213 seems to be a 50ohm resistor, so 25/50 = 0.5Amps through it giving 7.5W but it's not switched out. Measuring R213 in circuit shows 342ohms.. So something isn't right and I think I need to take the circuit board out and have a look.
Removing both fuses shows the transformer is fine still, it's delivering 104V peak-to-peak unloaded on both sides of the secondary winding.
So something is seriously up with the amp.. None of the FETs show outwards signs such as smoke, crack or discolouring (even of the dust). Not surprised after 24 years with the cooking high temps the amp runs at (hotter than you can touch on the top) that something has shorted.. well at least it gives me something to investigate/fix/rip&upgrade.
Last edited:
So investigating - this is the first time I've ever had the PCB out to look at the circuit on the underside.
The AC1-Positive diode in the bridge rectifier has gone to straight short. The BR is a 400V 10A piece with a heatsink. Although it's documented as a 200A surge rated for one cycle. It's got a large heatsink too so I suspect it's used as a form of current limit on start up too.
The 15000uF caps are tied directly to the BR without any soft start.. and so are the four 3300uF and two 470uF power rail caps plus a myriad of 1000uF and 100uF not including the foil caps. So in all there's 47,000+uF tied to the BR (some through transistors as regulators). So I suspect that given it failed on startup.. the BR finally had enough.
Once blown, the AC would then drag the caps all over the place. The 15000uF reserve filter cap at C12 for the positive rail that blew it's fuse out of circuit keeps a 13.6V charge for at least two hours and overnight it's only down to 15.5V from internal leakage.
So the area that I have a hot PCB trace is actually a mute for the opamp that switches the signal between the opamp and the power amps. So the mute doesn't work on the speakers (which means any power failure of the power amp devices results in full rail to the speakers).
The second N FET in from the right that shows darkened PCB tests good when out of circuit. I will test the others out of circuit some time today.
All the diodes test ok in circuit, the caps seem to block DC but I will test the larger caps out of circuit. I'd assume 6.5A would discolour the PCB but it appears not.
I'll take updates offline from this thread but just to give an update
The AC1-Positive diode in the bridge rectifier has gone to straight short. The BR is a 400V 10A piece with a heatsink. Although it's documented as a 200A surge rated for one cycle. It's got a large heatsink too so I suspect it's used as a form of current limit on start up too.
The 15000uF caps are tied directly to the BR without any soft start.. and so are the four 3300uF and two 470uF power rail caps plus a myriad of 1000uF and 100uF not including the foil caps. So in all there's 47,000+uF tied to the BR (some through transistors as regulators). So I suspect that given it failed on startup.. the BR finally had enough.
Once blown, the AC would then drag the caps all over the place. The 15000uF reserve filter cap at C12 for the positive rail that blew it's fuse out of circuit keeps a 13.6V charge for at least two hours and overnight it's only down to 15.5V from internal leakage.
So the area that I have a hot PCB trace is actually a mute for the opamp that switches the signal between the opamp and the power amps. So the mute doesn't work on the speakers (which means any power failure of the power amp devices results in full rail to the speakers).
The second N FET in from the right that shows darkened PCB tests good when out of circuit. I will test the others out of circuit some time today.
All the diodes test ok in circuit, the caps seem to block DC but I will test the larger caps out of circuit. I'd assume 6.5A would discolour the PCB but it appears not.
I'll take updates offline from this thread but just to give an update
@vishalk - when you tested your FETs did you test them controlling the gates?
My BUZ900P N-ch fet out of channel test results depended on if the gate was shorted or not. If I left them alone then they measure about 9ohms which is what you have written down. I suspect in my case the mosfets are ok, just the BR finally let go and possibly the caps took the hit. Annoying both my cheap multimeters need a replacement - eyeing yup a Bryman 387s when they arrive back in stock.
I'd agree with Liu735 - any hum the first thing to look at is the caps.
My BUZ900P N-ch fet out of channel test results depended on if the gate was shorted or not. If I left them alone then they measure about 9ohms which is what you have written down. I suspect in my case the mosfets are ok, just the BR finally let go and possibly the caps took the hit. Annoying both my cheap multimeters need a replacement - eyeing yup a Bryman 387s when they arrive back in stock.
I'd agree with Liu735 - any hum the first thing to look at is the caps.
Transformer burned A220, looking for a transformer made in China, the output voltage of 30V * 2, after replacing the sound is not what I expected, very disappointed, the quiescent current temporarily adjusted to about 150ma (single MOSFET). Fear of damage to the expensive speaker output added 2 relays. I can buy musical-fidelity E95/E300 use the transformer, the output 34V * 2, I wonder if it can be used?
What release of board is that?
Some differences I note between your board and my issue 3 board:
- No musical fidelity marking in the corner for the issue 3
- "video" instead of Tape 2 on the issue 3
- "preout" instead of tape out on the issue 3
- The VR1 volume control is mounted direct to the board and the single hold is on the other side compared to the issue 3 which sits on a set of wires.
- your gate resistors and bias resistors are different values to the issue 3.
- R1 (almost centre of the photo) is a different value to the issue 3
- R203 next to the photo supply is a large resistor on your board, on the issue 3 that's a small resistor - I guess that they had problems with overheating
- R209 mute delay in the photo is a different value
Interesting that you've subbed a 30V*2 into it. The original transformer I have is a 240Vac mains to 50-0-50 centre tap secondary 350Watts consumption statement on the back so I assume a 350W which is about the limit you want before inrush becomes an issue. I've been researching winding of toroidal transformers in detail (I want to wind my own for the valve amp) - is your winding two separate secondary windings or a centre-tapped single winding? What's the variation of resistance? The reason I ask is that two separate secondaries tend to not to be matched as precisely as centre tap (I don't suspect that MF haven't specced their toroid with bifilar windings for the CT but that could add a level of capacitance as a filter). Also you may want to check if your transformer for electrostatic band between the primary and secondary. The issue 3 transformer doesn't seem to have a outer magnetic GOSS band fitted.
Interesting that you have 30Vac, that gives 30*1.414 = 42Vpeak on each rail. The issue 3 having 50Vac peak would give ~70V and the caps are rated at 50V.. so something doesn't seem right with mine..
Last edited:
Can you confirm that the transformer output voltage is 50-0-50? If it is really so high, musical-fidelity is too undisciplined. diyaudio on someone said that the DC voltage is 42V, which indicates that the product is more confusing in various periods, so the failure rate is higher.
Last edited:
So my transformer:
Primary 230Vac ~22 ohm measured = ~4A (4*4*22 = 352Watts which is close enough to the "350W power consumption" on the back of the amp)
Secondary 0.8 ohm across entire secondary.
We can use the ratio sqr(Rs/Rp) and then ratio * 230Vac gives our secondary ac voltage. We don't know enough about the core (Al etc) so this gives:
sqr(0.8/22) * 230 = 43.86Vac or 43.86*1.414 = 62.02 Vdc full bridge rectified. Interesting given the caps are rated as 50Vdc..
Looking at a 300VA 2x55V tenma toroid data sheet the secondaries are 0.8141 ohm each. (0300P1_0209.pdf) unloaded is 4V higher than loaded which similar to my observation of 100Vpp loaded 104Vpp but I can't shake the feeling something is shorted.. There's full isolation between the mains and the secondary but the secondary was providing over 6.3A inrush to blow the fuse.
Either way with a dead short on the BR the caps on the V+ rail would have been subjected to full secondary +V to -V in the polar opposite which I assume means the caps are dead. At some point the voltage probably ended up as 100V as the cap kept the voltage at 50V whilst the AC wave dropped to -50V causing a massive current - all at inrush time and it's like charging a cap then reversing the charging cables..
My FETs seem ok - depending on the gate they can read infinite to low DCR. What I did find is that the FET with the circuit board discolouration seems to only be lightly thermally coupled to the heatsink compared to it's parallel companion that was almost stuck to it.
Course of action now is to replace the main caps that make up the V+ and V- rails, that's the 2x15000uF 50V, the 4x3300uF 50V and the two 470uF 35V.
What I'm also concerned about is that we don't know the regulation of the transformer. Given that we can have ±10% mains and then the transformer can add ±7% regulation for example, the 230V ac line could be 253Vac, sqr(0.8/22)*253 *1.414 = 68.23 Vdc then 7% upswing would result in 73Vdc. Something does seem odd and probably either my calculation or my two multimeters..
Primary 230Vac ~22 ohm measured = ~4A (4*4*22 = 352Watts which is close enough to the "350W power consumption" on the back of the amp)
Secondary 0.8 ohm across entire secondary.
We can use the ratio sqr(Rs/Rp) and then ratio * 230Vac gives our secondary ac voltage. We don't know enough about the core (Al etc) so this gives:
sqr(0.8/22) * 230 = 43.86Vac or 43.86*1.414 = 62.02 Vdc full bridge rectified. Interesting given the caps are rated as 50Vdc..
Looking at a 300VA 2x55V tenma toroid data sheet the secondaries are 0.8141 ohm each. (0300P1_0209.pdf) unloaded is 4V higher than loaded which similar to my observation of 100Vpp loaded 104Vpp but I can't shake the feeling something is shorted.. There's full isolation between the mains and the secondary but the secondary was providing over 6.3A inrush to blow the fuse.
Either way with a dead short on the BR the caps on the V+ rail would have been subjected to full secondary +V to -V in the polar opposite which I assume means the caps are dead. At some point the voltage probably ended up as 100V as the cap kept the voltage at 50V whilst the AC wave dropped to -50V causing a massive current - all at inrush time and it's like charging a cap then reversing the charging cables..
My FETs seem ok - depending on the gate they can read infinite to low DCR. What I did find is that the FET with the circuit board discolouration seems to only be lightly thermally coupled to the heatsink compared to it's parallel companion that was almost stuck to it.
Course of action now is to replace the main caps that make up the V+ and V- rails, that's the 2x15000uF 50V, the 4x3300uF 50V and the two 470uF 35V.
What I'm also concerned about is that we don't know the regulation of the transformer. Given that we can have ±10% mains and then the transformer can add ±7% regulation for example, the 230V ac line could be 253Vac, sqr(0.8/22)*253 *1.414 = 68.23 Vdc then 7% upswing would result in 73Vdc. Something does seem odd and probably either my calculation or my two multimeters..
I will check with a multimeter. I just remember seeing at the time the scope showing 100Vpp across the secondary (using two channels and no grounding clip).
Update - unloaded voltages:
MM1 measured 243Vac on the primary and 70Vac across the secondaries, centre to secondaries each 35Vac
MM2 measured 242Vac on the primary and 70Vac across the secondaries, centre to secondaries each 35Vac
So I suspect that both my multimeters and the user can't measured DCR of a toroidal accurately enough.
Seems odd though that I was getting a very low DCR on measuring the secondary.. Maybe when the caps were in place with the fuses caused issues when the scope was measuring, possibly the result f the failed BR.
So with that I suspect you're correct that it's a 30-0-30 transformer.
The secondaries still measure 0.8ohm end to end and 0.4 ohm CT-end. Perhaps they specifically used a low impedance secondary for higher current delivery.
So using 0.4 ohms (doh!):
sqr(0.4/22) * 230 = 31.1Vac, *1.414 = 44Vdc rails which seems closer to the measurements 242 line gives the amp 46Vdc rails.
With our +10% 253Vac line and +7% that could be as high as 36.5Vac, or 51.8Vdc when rectified. When the BR died the caps would have seen a polar opposite thus twice the voltage at 102Vdc). That matches close to the values I was seeing and would explain it..
MM1 measured 243Vac on the primary and 70Vac across the secondaries, centre to secondaries each 35Vac
MM2 measured 242Vac on the primary and 70Vac across the secondaries, centre to secondaries each 35Vac
So I suspect that both my multimeters and the user can't measured DCR of a toroidal accurately enough.
Seems odd though that I was getting a very low DCR on measuring the secondary.. Maybe when the caps were in place with the fuses caused issues when the scope was measuring, possibly the result f the failed BR.
So with that I suspect you're correct that it's a 30-0-30 transformer.
The secondaries still measure 0.8ohm end to end and 0.4 ohm CT-end. Perhaps they specifically used a low impedance secondary for higher current delivery.
So using 0.4 ohms (doh!):
sqr(0.4/22) * 230 = 31.1Vac, *1.414 = 44Vdc rails which seems closer to the measurements 242 line gives the amp 46Vdc rails.
With our +10% 253Vac line and +7% that could be as high as 36.5Vac, or 51.8Vdc when rectified. When the BR died the caps would have seen a polar opposite thus twice the voltage at 102Vdc). That matches close to the values I was seeing and would explain it..
Last edited:
I did find this thread: https://www.diyaudio.com/community/...nt-current-bias-problem-right-channel.148589/
With this schematic:
I've had a quick comparison with my A220 - this shows the Left side, the component identifiers match my board however there are some differences - firstly the rails aren't ±27Vdc (well I assume) and the caps show 35V or lower. The caps such as C12 (schematic bottom is +Vrail) is 50V as too are C111 and even C101. Yes it appears that the signal goes though a Jamicon 50V 0.22uF electrolytic.. Also the FETs aren MLP but BUZ900P and BUZ905P for the issue 3 board.
I'll compare this more tonight and I may even make up a LTSpice model.
With this schematic:
I've had a quick comparison with my A220 - this shows the Left side, the component identifiers match my board however there are some differences - firstly the rails aren't ±27Vdc (well I assume) and the caps show 35V or lower. The caps such as C12 (schematic bottom is +Vrail) is 50V as too are C111 and even C101. Yes it appears that the signal goes though a Jamicon 50V 0.22uF electrolytic.. Also the FETs aren MLP but BUZ900P and BUZ905P for the issue 3 board.
I'll compare this more tonight and I may even make up a LTSpice model.