Leinster Lad Posted November 11, 2020 Share Posted November 11, 2020 If I warm up Q1 with my finger, the offset voltage drops. When I remove my finger it goes back up. Doing the same to Q2 has the opposite effect Link to comment Share on other sites More sharing options...
MarcAL Posted November 11, 2020 Author Share Posted November 11, 2020 That makes sense. You will have to wait till they warm up and then hopefully they will remain stable. With a bit of tweaking you may get a much lower DC offset. Mine are between -20mVDC and -24mVDC. Link to comment Share on other sites More sharing options...
Leinster Lad Posted November 11, 2020 Share Posted November 11, 2020 Normally Q1 and Q2 (input differential pair) are thermally bonded together. On this amp they are too far apart for that. The text mentions something about something that means that they don't even have to be matched ???? confusing. Tilbrook sometime frequents the diyAudio forum, I might ask the question over there Link to comment Share on other sites More sharing options...
Leinster Lad Posted November 11, 2020 Share Posted November 11, 2020 I could try this arrangement that is used in some amps to adjust the offset to zero. Perreaux use a very similar arrangement. Couldn't hurt to tinker, right ? You can see the voltage divider connected to the input. No idea is this will affect the sound quality ?? Link to comment Share on other sites More sharing options...
Leinster Lad Posted November 11, 2020 Share Posted November 11, 2020 4 hours ago, andyr said: 70mV? Won't do any harm but it would be noice to get it down to, say +/-10mV? My (very limited!) understanding is that it's a resistor value which sets the offset. In the case of the amps which I am currently building, the pcb designer used a very neat trick! He used a connector in parallel with a pair of holes which take the final resistor. You first of all insert the 3 legs of a Bourne pot into the connector ... adjust the pot for zero offset - and then remove it and insert the final resistor having the pot value. Andy Hey Andy, thanks. What amp has this "feature" that you mentioned and do you have a copy of the schematic ? cheers Dave Link to comment Share on other sites More sharing options...
andyr Posted November 11, 2020 Share Posted November 11, 2020 8 hours ago, Leinster Lad said: Hey Andy, thanks. What amp has this "feature" that you mentioned and do you have a copy of the schematic ? cheers Dave Here's a link, Dave: https://www.diyaudio.com/forums/solid-state/344540-alpha-nirvana-39w-8ohm-class-amp.html The circuit was designed by Hugh Dean ... others did the PCB layout/production. The area around R4 (on the 2nd schematic) is where the DC offset is set. PCBs are available - plus a matching PS PCB. Andy Link to comment Share on other sites More sharing options...
Leinster Lad Posted November 20, 2020 Share Posted November 20, 2020 Well I tried the pot on the emitters of the dif pair. Did nothing until the pot wiper reached the end of the resistance track and effectively shorted out that sides emitter resistor. I have tried very closely matching both hfe and Vbe, which reduced the offset from the original 80mV down to about 50mV. I'm going to bite the bullet and recap it. Then I'll replace the BF's with KSA1142/KSC2682 once I've got the heat sink fabrication done. I'll do the 18V zener mod at the same time. The "0" volt star point is not exactly a "star" , so will get that sorted just in case it is creating some weird voltage offset which is being picked up at the input. Link to comment Share on other sites More sharing options...
MarcAL Posted November 20, 2020 Author Share Posted November 20, 2020 Hopefully correcting the star point should get rid of some unwanted noise. What happens if you change R8 from 10K to 22K? Link to comment Share on other sites More sharing options...
Leinster Lad Posted November 20, 2020 Share Posted November 20, 2020 1 hour ago, MarcAL said: Hopefully correcting the star point should get rid of some unwanted noise. What happens if you change R8 from 10K to 22K? Noise ? haven't got that far yet. Still trying to sort out the offset voltage. Have not touched R8 yet. Time is a bit of an issue at the moment. Link to comment Share on other sites More sharing options...
MarcAL Posted November 20, 2020 Author Share Posted November 20, 2020 Do you have much noise? Sometimes by fixing one problem it can fix another at the same time. Link to comment Share on other sites More sharing options...
Leinster Lad Posted November 21, 2020 Share Posted November 21, 2020 Noise? No idea, have not even connected any speakers to it yet ! Hum? Hiss? distortion? I might have a bit more of an idea once (if) I can get my scope on it, and as I'm not trained in these things, will have to figure out what it tells me. Meaningful sized blocks of time are few and far between at the moment, so this is going to take a while. Link to comment Share on other sites More sharing options...
Leinster Lad Posted November 25, 2020 Share Posted November 25, 2020 A bit more poking around and found approx 47.4 volts across the emitter/collector of Q1 and Q2 (BC550 ) These are only rated for 45Vceo, so I guess they are on the point of ce breakdown. Whether this is contributing to the offset I don't know, but it is never wise to run and component at or above its maximum rated spec. I've ordered some 20 x BC546 (element14) (rated to 65Vceo) which I will match up a pair and try. I'll up R8 to 22K and see what happens. It has also been suggested that I could also decrease R2 down to ~ 18K (from 47K) should help to equalise the bias across Q1 and Q2 (diff pair) Not really sure how "off spec" I really want to go with this amp. I am just trying to get the offset voltage down to what the designer said would be normal.(25mV) Link to comment Share on other sites More sharing options...
Monkeyboi Posted November 25, 2020 Share Posted November 25, 2020 56 minutes ago, Leinster Lad said: A bit more poking around and found approx 47.4 volts across the emitter/collector of Q1 and Q2 (BC550 ) These are only rated for 45Vceo, so I guess they are on the point of ce breakdown. Whether this is contributing to the offset I don't know, but it is never wise to run and component at or above its maximum rated spec. I've ordered some 20 x BC546 (element14) (rated to 65Vceo) which I will match up a pair and try. I'll up R8 to 22K and see what happens. It has also been suggested that I could also decrease R2 down to ~ 18K (from 47K) should help to equalise the bias across Q1 and Q2 (diff pair) Not really sure how "off spec" I really want to go with this amp. I am just trying to get the offset voltage down to what the designer said would be normal.(25mV) Nothing wrong with using transistors with equivalent gain and a higher spec'ed Vceo in this case. Before going off on a tangent is the current flowing through Q13 correct at about the nominal 450uA? This can be quickly verified by measuring the volt drop across R11 (47k) . if correct the voltage should be about 11.28v based on an Ohm's Law calculation. Furthermore, you can verify if Q1 and Q2 are conducting equally by measuring the volt drop across their respective emitter resistors R10 for Q1 and R13 for Q2. Both R10 and R13 are 270R. If Q1 and Q2 are conducting equally as a differential pair. There should be very little voltage difference under no signal conditions between the collectors of Q1 and Q2. In direct coupled designs it's possible any other components are contributing to the small d.c. offset you are seeing at the output. IMHO 25mV isn't too much to worry about. As Q1 and Q2 drive Q3 and Q4, it's probably worth checking that Q3 and Q4 are sinking the nominal ~6mA of current equally between them. Verify this by measuring the volt drop across R7 and R18 (both 270R) to check that they are close to equal. # All these readings are nominal only and in practice may vary slightly due to production variations in components. The readings are based on the modified circuit with the 18v zener wired between the collectors of Q5 and Q7. Cheers, Alan R. Link to comment Share on other sites More sharing options...
Leinster Lad Posted November 25, 2020 Share Posted November 25, 2020 Thanks Alan, I did measure volt drops across those resistors that you mentioned when I first started investigating this offset voltage issue. From memory there was nothing of note to concern me. I will measure ( and record ) these figures again once I've replaced the BC550's with BC546's The offset that I have is 53mV and 62mV. It was around the 80mV range when I first got it. The 25mV is what DT specified would be realistic to achieve from the kit. I have had a bit of a look for alternative transistors for the diff pair, and the BC546 are the higher voltage rated equivalent of the BC550. Link to comment Share on other sites More sharing options...
Monkeyboi Posted November 25, 2020 Share Posted November 25, 2020 3 hours ago, Leinster Lad said: Thanks Alan, I did measure volt drops across those resistors that you mentioned when I first started investigating this offset voltage issue. From memory there was nothing of note to concern me. I will measure ( and record ) these figures again once I've replaced the BC550's with BC546's The offset that I have is 53mV and 62mV. It was around the 80mV range when I first got it. The 25mV is what DT specified would be realistic to achieve from the kit. I have had a bit of a look for alternative transistors for the diff pair, and the BC546 are the higher voltage rated equivalent of the BC550. Looks like you've got the bases covered. The only way to reduce the d.c. offset IMHO is to match the gains of the transistors as closely as can practically achieved. Yes,it would be nice to get it as low as possible, however Tilbrook never allowed for any offset adjustment in his design. I guess you could try compensating by attempting to match the current flow differences between Q1 and Q2. Several methods other than exact gain matching could be tried. How practical it might be is another issue. Cheers, Alan R. Link to comment Share on other sites More sharing options...
Leinster Lad Posted November 25, 2020 Share Posted November 25, 2020 Did that and tried the other. Matched both hfe and Vbe to within a poofteenth of each other. hfe within 1 and Vbe exactly the same. Even with waiting for temp stabilisation. Managed to get the offset down from around 80mV to around 50mV The trimpot idea did basically nothing until the wiper dropped off the end of the track and went to zero ohms. That saw the offset drop to 18mV Link to comment Share on other sites More sharing options...
Leinster Lad Posted November 27, 2020 Share Posted November 27, 2020 On 25/11/2020 at 10:40 AM, Monkeyboi said: This can be quickly verified by measuring the volt drop across R11 (47k) . if correct the voltage should be about 11.28v based on an Ohm's Law calculation. L R R11, 10.67V and 10.88V On 25/11/2020 at 10:40 AM, Monkeyboi said: R10 for Q1 and R13 for Q2 L R R10, 30.9V & 31.3V R13, 30.5V & 30.8V On 25/11/2020 at 10:40 AM, Monkeyboi said: worth checking that Q3 and Q4 are sinking the nominal ~6mA of current equally between them. Verify this by measuring the volt drop across R7 and R18 (both 270R) to check that they are close to equal. L R R7, 1.61V & 1.60V R18, 1.64V & 1.63V So, close but there is a bit of difference. I'm going to wait until the BC546's arrive and fit them. Then I'll check the voltages again. Measured the Vce on the BC550's installed and Q1 & Q2 are all around 49.5V = way too high, possibly causing a bit of breakdown ? (max = 45V) Q3 is acceptable at 38.0V, but i'll replace it with a 546 anyway. 1 Link to comment Share on other sites More sharing options...
Leinster Lad Posted December 1, 2020 Share Posted December 1, 2020 Just having a play with the scope, looking at the pwr supplies. Roughly 55mV RMS on each rail with about 180mV P-P. I also check my “rebuild” series 5000 and that was 35mV and 130mV P-P Link to comment Share on other sites More sharing options...
Leinster Lad Posted December 1, 2020 Share Posted December 1, 2020 Also built a dummy load, switchable between 50 and 100 ohms ( 2 channels) to load test the pwr supply. Got a fan for cooling as well. @ 50 volts, it pulls 0.5 or 1 amp 2 Link to comment Share on other sites More sharing options...
Guest Suzyj Posted December 2, 2020 Share Posted December 2, 2020 On 15/08/2020 at 7:40 PM, Leinster Lad said: Yep, they certainly did rip Suzy off. From what I have read, she designed and developed an absolutely cracking amp. Unfortunately, she is not (able?) to support her work and sell boards etc. These connex guys have just filled a vacuum created by that. Sad really. I'd by a pair of her boards in a heart beat. That Suzy woman is a right royal pain, I reckon. My old AEM6000 derivatives are still going strong after rather a lot of use. I’m more than happy to support the design. One of these days I’ll redraw the current versions in kicad so they’re a bit more accessible than the Protel ones. The 477 is a very nice amp. It has the benefit of not needing hard-to-find jfets for the front end, too. Link to comment Share on other sites More sharing options...
Leinster Lad Posted December 2, 2020 Share Posted December 2, 2020 19 minutes ago, Suzyj said: That Suzy woman is a right royal pain, I reckon. Pretty sure she is not Link to comment Share on other sites More sharing options...
Guest Suzyj Posted December 2, 2020 Share Posted December 2, 2020 I reckon you’ll get more joy adding Zeners (say 15V or thereabouts) in series with the input stage collectors than just replacing them with higher voltage parts. It’ll reduce the power dissipated by the input transistors, making them run cooler etc. Otherwise use a cascode stage in the same spot. Link to comment Share on other sites More sharing options...
Leinster Lad Posted December 21, 2020 Share Posted December 21, 2020 Hey folks a bit of an update. Substituted the bc550’a for bc546’s (higher voltage rating) , matched them real good ( ON semi from element14). And they made no difference to the output offset voltage. So now to the VAS. I’ve got some 2sa1142 and 2sc2682 which some folk over on diyaudio recommended as substitutes. Matched them up to within a bees dick. Currently making suitable heatsink for them. Lots of drilling and tapping and measuring and filing. Should work a treat ! Link to comment Share on other sites More sharing options...
Leinster Lad Posted December 21, 2020 Share Posted December 21, 2020 this is how it will look when on the board. About 10 mm clearance to the top of the case 1 Link to comment Share on other sites More sharing options...
MarcAL Posted March 22, 2021 Author Share Posted March 22, 2021 How has it been progressing LL? Link to comment Share on other sites More sharing options...
Recommended Posts