Hmm, perhaps C208 could have been damaged?
Merlin, in his power supplies book, says R208, in combination with D209, "provides and essential soft current limit". D209 protects the FET from inrush at startup and 'careless probing'. So R208 shouldn't have any impact on hum levels. Do you have a scope (and suitable probes, 100x?) to measure the ripple before and after the FET? You could test at the heater elevation point too - that would be a scaled version of the HT ripple.
Going back up the chain - the rectifier diodes should have plenty of capacity so can't think any of them would be damaged. TH201 shouldn't have any impact - it's designed to get hot. That leaves C205 and C210.
Merlin, in his power supplies book, says R208, in combination with D209, "provides and essential soft current limit". D209 protects the FET from inrush at startup and 'careless probing'. So R208 shouldn't have any impact on hum levels. Do you have a scope (and suitable probes, 100x?) to measure the ripple before and after the FET? You could test at the heater elevation point too - that would be a scaled version of the HT ripple.
Going back up the chain - the rectifier diodes should have plenty of capacity so can't think any of them would be damaged. TH201 shouldn't have any impact - it's designed to get hot. That leaves C205 and C210.
Thanks for the easy access schematic. I wonder if @tristanc could explain the operation of his high voltage ripple filter design compared to Pete’s. Pete’s I understand, but have some difficulty with the new one. I guess tristanc followed a design by Merlin, but I don’t have his PS book.
One observation is that tristan’s did not use a 1k gate stopper resistor that Pete used, whereas every ripple filter/capacitance multiplier I have seen featured a gate stopper > 220 to suppress oscillation of the Mosfet. Could that be an issue and caused the high temperature on the Mosfet that destructed?
I am still thinking about the possible causes for the increased hum you are observing.
Last edited:
I guess R206 could be increased to 1k as another belt and braces step.
The two forms of this capacitor multiplier, as I understand it, are essentially the same - just drawn slightly different. Pete leaves out R208. Yes, slight variation in values, but nothing that would cause a change in operation.
An oscillation could potentially cause the destruction. But that wouldn't explain the hum now - unless it's oscillating at a differing frequency.
To recap, 2 sensible steps would be to increase the power rating of R208 to >=2W & to increase R206 to 1k. Whilst you were at it, R207 & R209 could always be given the same treatment.
I'll try and make the same changes when I have the amp on the bench.
The two forms of this capacitor multiplier, as I understand it, are essentially the same - just drawn slightly different. Pete leaves out R208. Yes, slight variation in values, but nothing that would cause a change in operation.
An oscillation could potentially cause the destruction. But that wouldn't explain the hum now - unless it's oscillating at a differing frequency.
To recap, 2 sensible steps would be to increase the power rating of R208 to >=2W & to increase R206 to 1k. Whilst you were at it, R207 & R209 could always be given the same treatment.
I'll try and make the same changes when I have the amp on the bench.
Yep - I did a very quick eyeball of the two datasheets and thought they looked pretty darn similar. We're not missing something obvious like the pins being in a different order? The original BH PCBs had something like that and people didn't notice the non-functioning filter. ie, the RCs were doing fine on their own.
Like this:
It's the same heatsink as in the BOM.
I'll order beefier resistors. I think I have a replacement diode.
Merlin's schematic (with ~310VDC input and ~300VDC output) shows a STP11NK40ZFP being used. These are available at Mouser with the same pin out, albeit 400V parts. To my untrained eyes, the part you have used is pretty close to both this and the part I used.
It has been a while, but I think I went through the calculations for selecting a suitable heatsink.
It has been a while, but I think I went through the calculations for selecting a suitable heatsink.
I have been pondering the Safe Operating Area (SOA) of these Mosfets, which I believe are not primarily intended for DC, but rather pulse currents. What is the actual quiescent voltage drop over the Mosfet? Since @tristanc mentioned a 10Vdc over the Mosfet, here are my non-expert observations and thoughts.
The spec sheets (https://www.mouser.com/datasheet/2/389/std5n62k3-1850498.pdf) fig 5 shows a maximum of about 170 ma at a Vds drop of 10V, if I read it correctly. But it also states the current is for a single pulse, and interpolating from the increase in pulse duration shown on the graph, I doubt that 170 ma could be sustained as DC.
Comparing the SOA for STP11NK40ZFP it appears better than STF5N62K3 wrt this constraint, able to handle -700 ma at 10V drop (at the relevant pulse current), so it might be good at 200-300 ma DC.
The heatsink looks smaller than expected, but might be fine if calculated for adequate dissipation.
As failure mechanism, I’m not sure what failed first, the Mosfet or R208. At full power the current draw from the HV PS must >=220 ma, and that causes R208 to dissipate > 1 watt. It is interesting that R208 lasted as log as it did.
The spec sheets (https://www.mouser.com/datasheet/2/389/std5n62k3-1850498.pdf) fig 5 shows a maximum of about 170 ma at a Vds drop of 10V, if I read it correctly. But it also states the current is for a single pulse, and interpolating from the increase in pulse duration shown on the graph, I doubt that 170 ma could be sustained as DC.
Comparing the SOA for STP11NK40ZFP it appears better than STF5N62K3 wrt this constraint, able to handle -700 ma at 10V drop (at the relevant pulse current), so it might be good at 200-300 ma DC.
The heatsink looks smaller than expected, but might be fine if calculated for adequate dissipation.
As failure mechanism, I’m not sure what failed first, the Mosfet or R208. At full power the current draw from the HV PS must >=220 ma, and that causes R208 to dissipate > 1 watt. It is interesting that R208 lasted as log as it did.
Last edited:
And the 0.6 watt R208 is still going in @tristanc’s build.
As for the increased hum after the failure of the mosfet, that is what I would expect with sensitive speakers and ears in this (relatively) high current situation where the HV capacitance multiplier no longer multiplies. The error on the “original PCB’s” that went unnoticed for a while was actually in the -135 V bias supply with very little current draw.
As for the increased hum after the failure of the mosfet, that is what I would expect with sensitive speakers and ears in this (relatively) high current situation where the HV capacitance multiplier no longer multiplies. The error on the “original PCB’s” that went unnoticed for a while was actually in the -135 V bias supply with very little current draw.
Last edited:
A couple of discussions / posts from the forum of interest:
https://www.diyaudio.com/community/...tance-multiplier-bypass-cap-necessary.381792/
And Pete joining a discussion on the original DCPP:
https://www.diyaudio.com/community/...p-power-amp-design.151206/page-4#post-1927445
And the last paragraph at https://sound-au.com/project15.htm
https://www.diyaudio.com/community/...tance-multiplier-bypass-cap-necessary.381792/
And Pete joining a discussion on the original DCPP:
https://www.diyaudio.com/community/...p-power-amp-design.151206/page-4#post-1927445
And the last paragraph at https://sound-au.com/project15.htm
Last edited:
Apologies, I misread the SOA graph. The graph shows 1.17 amp in stead of 170 ma as I mistakenly read. It looks indeed like the STD5N62K3 should be able to sustain the required 200ma for the BH circuit.
Last edited:
An update:
I tested the new STD5N62K3 I put in and it has failed too. Shorted between drain and source. Is it possible that it failed because of the dead diode? It might also be an explanation for the increased hum if the new STD5N62K3 failed immediately on power up.
My plan now:
I tested the new STD5N62K3 I put in and it has failed too. Shorted between drain and source. Is it possible that it failed because of the dead diode? It might also be an explanation for the increased hum if the new STD5N62K3 failed immediately on power up.
My plan now:
- Replace the failed STD5N62K3 with another STD5N62K3 (unless there is another MOSFET available that is more suitable)
- Replace the failed diode D209
- Replace R208 and R209 with 2 W resistors
- Replace R206 and R207 with 1K resistors
I just stumbled over this nice project. Large kudos for sharing everything on github! I am working on publishing my latest project in the same way once it's getting there. One thing I noticed with the Baby Huey files is that Kicad can't find the footprints for the tubes (or tube sockets). I guess they are missing in the repository.
No idea I'm afraid, but it is very strange. TH201 is there to prevent inrush, but if R208 was jumpered perhaps that wasn't enough? Although in the original DCPP R208 isn't there... It's such a simple arrangement - in fact, perhaps one step would be to bypass the whole lot (go from C205 to C210 with a 2W 10k resistor, e.g., tacked in, if you have one around) and check the drop it, e.g..
Yes. I'll likely do the same for the resistors to be on the safe side - although how on earth mine has lasted almost 2 years!?
Thanks for the kind comment.
You may need to add the library (TristanValve.pretty) or re-link them. I forget the terminology. I upgraded KiCAD during the design and so many bits changed I may have failed to store the footprints correctly. I'll take another look.
That didn't go well. The parts arrived today and I soldered them in as planned in post #95 when I got home from work. I biased the tubes without any apparent issues, but when I connected it up to my speakers there was a hum a bit loader than the last time. It seems that both the MOSFET and the diode has failed like last time.
I'm at a loss about what to do or check. Any suggestions?
I'm at a loss about what to do or check. Any suggestions?
Jeepers. All I can think is to bypass the whole section with a 10k 2W resistor (or any value as long as it can take the power) and check things operate normally. This would rule out a short on, e.g. socket pins. Given it did work fine for a while this is unlikely. But you never know.
2nd would be replacing all the electrolytic caps around that area. The diode and FET would only burn up with increased current. It must go somewhere. If not through the valves there may be another route.
Also, perhaps check if operation a channel at a time. I.e, if you bypass the regulator only use channel 1 with valves installed, check for stable operation, sound etc. then repeat for channel 2. Measure the voltage drop across the temporary resistor. Check currents. It may point to a fault downstream of the regulator.
2nd would be replacing all the electrolytic caps around that area. The diode and FET would only burn up with increased current. It must go somewhere. If not through the valves there may be another route.
Also, perhaps check if operation a channel at a time. I.e, if you bypass the regulator only use channel 1 with valves installed, check for stable operation, sound etc. then repeat for channel 2. Measure the voltage drop across the temporary resistor. Check currents. It may point to a fault downstream of the regulator.