low resistance adapter
this is a pretty neat looking 4-wire low resistance DMM adapter...
http://www.angelfire.com/electronic/funwithtubes/low_res_meter.html
the test leads are dodgy but I'm sure they do the job 😉
this is a pretty neat looking 4-wire low resistance DMM adapter...
http://www.angelfire.com/electronic/funwithtubes/low_res_meter.html
the test leads are dodgy but I'm sure they do the job 😉
Thanks Nelson,
not able to touch the amp until tomorrow, but DC offset was about 45-50mv on the suspect monobloc which I think indicates that the 9610's are basically ok. I'm going to remove the driver board from the list of suspects for the moment.
I suspect at this stage that either the cheap silpads are causing problems, or one (or more) of the mosfets is faulty. I'll take a look at the resistors.
I've ordered some BQ silpad 2000 to replace the junky ones I have installed at present.
cheers
Paul
not able to touch the amp until tomorrow, but DC offset was about 45-50mv on the suspect monobloc which I think indicates that the 9610's are basically ok. I'm going to remove the driver board from the list of suspects for the moment.
I suspect at this stage that either the cheap silpads are causing problems, or one (or more) of the mosfets is faulty. I'll take a look at the resistors.
I've ordered some BQ silpad 2000 to replace the junky ones I have installed at present.
cheers
Paul
After taking a look at the archives I tried out a few tests on the output boards.
I measured D-S on the mosfets and all seemed to behave as expected, giving 0.832 in one direction and nothing in the other. Checked the other OP boards and got very similar readings.
I hooked up a 9V battery to V-/+ & C-/+ on both and checked voltage drop over source and gate resistors. There was a V drop across source resistors and zero voltage drop across the gate resistors. This should mean the 240's are ok.
I'm beginning to wonder if this isn't simply an issue of poor thermal transfer between some of the mosfets and the heatsink? Am I right in thinking the MOSFETS running the hottest will draw more current than a cooler MOSFET??
cheers
Paul
I measured D-S on the mosfets and all seemed to behave as expected, giving 0.832 in one direction and nothing in the other. Checked the other OP boards and got very similar readings.
I hooked up a 9V battery to V-/+ & C-/+ on both and checked voltage drop over source and gate resistors. There was a V drop across source resistors and zero voltage drop across the gate resistors. This should mean the 240's are ok.
I'm beginning to wonder if this isn't simply an issue of poor thermal transfer between some of the mosfets and the heatsink? Am I right in thinking the MOSFETS running the hottest will draw more current than a cooler MOSFET??
cheers
Paul
In the end we were both wrong Maousse.
the problem was a broken pin in the gnd terminal block on the psu pcb. and of course I had chassis gnd wired to that connector. PSU measured ok from the good pin to +/- but was floating against the chassis and driver/op boards.
V drop over source resistors is not perfect but ranges from 409mv to 360mv on the -tve op board. dc offset is 14.5mv.
the problem was a broken pin in the gnd terminal block on the psu pcb. and of course I had chassis gnd wired to that connector. PSU measured ok from the good pin to +/- but was floating against the chassis and driver/op boards.
V drop over source resistors is not perfect but ranges from 409mv to 360mv on the -tve op board. dc offset is 14.5mv.
Oh, glad you managed to find the problem.
You've got too strong arm for the poor little terminal screws!
You've got too strong arm for the poor little terminal screws!
unfortunately not!! the part that soldered into the board was fairly thin flat metal rather than a nice sturdy pin. Fortunately I had a replacement on hand.
I'll tackle checking the resistors in the next week or so, but the amps are sounding pretty sweet as they are so the temptation is to leave as is.
Just need to get the D1V3 hooked up now to complete the passified front-end.
I'll tackle checking the resistors in the next week or so, but the amps are sounding pretty sweet as they are so the temptation is to leave as is.
Just need to get the D1V3 hooked up now to complete the passified front-end.
Resistors that are cosmetically fine should be OK, we are not in high voltage here 😉
I'd rather check the semiconductors, just in case... but IRF are solid stuff 😛
I'd rather check the semiconductors, just in case... but IRF are solid stuff 😛
240's should be fine, I'm not sure what current level Tarasque matches at but the sets should be reasonably close as they were labeled within 0.001V. I meant double checking resistance values to get a better match on each board, but that will mean putting together a tester of some description - and pulling the amps apart again.
wouldn't an amp with IRFP240's typically run at more like 5-600ma+?? 
That certainly explain's why the voltage drop over source resistors is showing a relatively large spread. They are definitely not matched in the sense of "matched at close to operating conditions".
I may as well forget about fiddling with resistors until I find time to properly match a set MOSFETS.
That certainly explain's why the voltage drop over source resistors is showing a relatively large spread. They are definitely not matched in the sense of "matched at close to operating conditions".
I may as well forget about fiddling with resistors until I find time to properly match a set MOSFETS.
Don't under estimate the influence of teh source resistors and the different thermal aspects for each fet when mounted. When you do not use the same force to mount each fet, they almost certanly will have different properties.
Also matching at high currents decreases the accuracy of teh match, as the junction temperature then is even more critical.
Also the mounting aspects on the heatsink becomes an issue.
I build an X1000 alike and would allow 5% spread in idle state between the fets (7,5W per fet). I start with a batch of about 600 IRFP240 and 600 IRFP9240. From these I select 2 sets of 50 fets as close as possible at the approx idle current. I mount the heatsink assembly and bring it to approx idle operating condition. I measure each fet and every one outside of teh 5% band I replace with one of teh remaining 10 fets until it is OK. Prior to mounting the source resistors I measure them so I have 40 resistors within 0,1%.
You can do a similar thing with the A30, although I did not hear a difference between a channel matched (in circuit) to 1% and one that was within 30%.
I have to say that according to reliability calculations, a channel that has fets with similar dissipation will survive longer than one with 30% spread. (~10 years instead of 15 years)
I'm not worried about that as I keep building stuff anyway just for the fun!
Also matching at high currents decreases the accuracy of teh match, as the junction temperature then is even more critical.
Also the mounting aspects on the heatsink becomes an issue.
I build an X1000 alike and would allow 5% spread in idle state between the fets (7,5W per fet). I start with a batch of about 600 IRFP240 and 600 IRFP9240. From these I select 2 sets of 50 fets as close as possible at the approx idle current. I mount the heatsink assembly and bring it to approx idle operating condition. I measure each fet and every one outside of teh 5% band I replace with one of teh remaining 10 fets until it is OK. Prior to mounting the source resistors I measure them so I have 40 resistors within 0,1%.
You can do a similar thing with the A30, although I did not hear a difference between a channel matched (in circuit) to 1% and one that was within 30%.
I have to say that according to reliability calculations, a channel that has fets with similar dissipation will survive longer than one with 30% spread. (~10 years instead of 15 years)
I'm not worried about that as I keep building stuff anyway just for the fun!
Hi Tarasque,
Thanks for the considered reply. I did some searching and found a post from Nelson in which he opined that the main advantage of matched FETS is for long term reliability, and that there is no real audible benefit, which agrees with your findings.
Interesting points regarding mounting. I have seen mention of NASA requiring use of calibrated torque wrenches to ensure that all devices are mounted consistently and I suppose it is for this very reason.
I also take your point on source resistor matching and will try to get these other factors matched before looking at the 240's.
NP mentions in an article measuring VGS at 200mA as you do, but also talks about an additional test at 0.5A to check for similar transconductance, so I would assume that FETs matched on both parameters would be closer to ideal, and would run closer matched at operating conditions. I'm sure the yield from matching like this would be extremely low and the process time consuming.
In a thread on "10 was to improve your Aleph" NP mentions as one of his top tips matching FETs for linearity to reduce distortion. I'd guess that sets matched for VGS and transconductance would be more linear than those matched at one point.
Anyway bottom line is that amp is sounding pretty fantastic as is. The amp was out of action for so long that I can't make any real assessment of the difference of the changes I made, but I think I hear slightly more detail and slightly better imaging than previously.
Summary of parts changes:
Replace IRFP044's with matched IRFP240
Replace all 0.47R 5% vishay ww's with 1% Mills MRA-5
Replace 220uf/35V nichicon VZ with rubycon ZA
Replace css zener bypass cap with 0.47uF MKP1837 film cap
Bypass input electrolytic with 0.01uf MKP1837 polyprop film cap
Replace signal and power wiring with kimber kable hookup wire
Thanks for the considered reply. I did some searching and found a post from Nelson in which he opined that the main advantage of matched FETS is for long term reliability, and that there is no real audible benefit, which agrees with your findings.
Interesting points regarding mounting. I have seen mention of NASA requiring use of calibrated torque wrenches to ensure that all devices are mounted consistently and I suppose it is for this very reason.
I also take your point on source resistor matching and will try to get these other factors matched before looking at the 240's.
NP mentions in an article measuring VGS at 200mA as you do, but also talks about an additional test at 0.5A to check for similar transconductance, so I would assume that FETs matched on both parameters would be closer to ideal, and would run closer matched at operating conditions. I'm sure the yield from matching like this would be extremely low and the process time consuming.
In a thread on "10 was to improve your Aleph" NP mentions as one of his top tips matching FETs for linearity to reduce distortion. I'd guess that sets matched for VGS and transconductance would be more linear than those matched at one point.
Anyway bottom line is that amp is sounding pretty fantastic as is. The amp was out of action for so long that I can't make any real assessment of the difference of the changes I made, but I think I hear slightly more detail and slightly better imaging than previously.
Summary of parts changes:
Replace IRFP044's with matched IRFP240
Replace all 0.47R 5% vishay ww's with 1% Mills MRA-5
Replace 220uf/35V nichicon VZ with rubycon ZA
Replace css zener bypass cap with 0.47uF MKP1837 film cap
Bypass input electrolytic with 0.01uf MKP1837 polyprop film cap
Replace signal and power wiring with kimber kable hookup wire
They sing very nicely. It's been a little warm (34degC) here yesterday and forecast warmer today and no aircon where I live, so I'm resting the amps. I might get a chance to listen again today before it gets too hot.
As I noted above it's a little hard to evaluate the changes given the 5-6 weeks of downtime since the amp meltdown. The A30's are definitely sounding very nice - they always did - but I think the changes have brought a slight improvement to imaging and detail. I have noticed details that I wasn't previously aware of in familiar recordings so there has to be something that has changed for the better... 😀
As I noted above it's a little hard to evaluate the changes given the 5-6 weeks of downtime since the amp meltdown. The A30's are definitely sounding very nice - they always did - but I think the changes have brought a slight improvement to imaging and detail. I have noticed details that I wasn't previously aware of in familiar recordings so there has to be something that has changed for the better... 😀
- Status
- Not open for further replies.