Only way that resistors cause distortion is that they are of low quality and maybe change their value when hot. Otherwise the reason for distortion is that you messed up something
When it comes to instabilities in DC offset and power off thump, you have to determine does it come from LTP or its' CCS.
Measure the current from input CCS and if it's stable the culprit is in LTP.
10-15mV offset variation is nothing to worry about. See if you get that thump when you remove 22uF over LED.
Last edited:
juma - sipi told me that you're nutz ....
off course he is nutz - listening to us, c'mon!
Only way that resistors cause distortion is that they are of low quality and maybe change their value when hot. Otherwise the reason for distortion is that you messed up something
The only reason I asked was because both schematics explicitly said 5W... In any event the distortion is almost certainly because I messed up something...
I had the thump before putting the 22uF in, so I'm sure that's not causing it. I'm going to get the source resistors changed to equal values and then reevealuate everything. (Including the LTP/CCS query you mention, if I can.)
(OT: I'm back from the "parental duties" I mentioned above - a visit to the Science and Technology fair the Brazilian government have running this week. Lots of fun things for 5-6 year-old kids and their dads... I really should have planned on spending the whole day... Oh well, that's for next year...)
Cheers
Nigel
Trouble and more trouble....
Well, I switched the source resistors R7 to match R16 (at 0R56) and connected everything up to test... To my annoyance the DC offset was way off, and worse, the LED on the right channel started smoking....
I've been checking things out for a couple of hours, and the voltages in the right channel are all wrong - total DC offset on the output is about +18V, that is, the rail voltage. I can't believe this all happened just from changing R7....
I need a little help thinking this through, here...
Collector of Q1 is about 0.43 - which I guess has to be corrected with the trimpot, but isn't so bad, so the CCS seems OK, I guess, now I've changed the LED. So that suggests something is wrong either in the Aleph current source or the output stage around Q3. Voltages across R7 and R16 are 0.628 and 0.611, so bias current is about 1.1A, which appears to be a little higher than it was before increasing R7 from 0R47 to 0R56, which seems odd. There are no obvious shorts, and I tried switching Q8 for a new BC557 to see ifthat was the problem, but no...
Is there an easy way of checking Q3 without taking it out of the circuit? Could there be something wrong with it other than an obvious short (which the DMM doesn't show...)?
I can pull a BD139 off an old board and switch Q1 out, if neessary, but it doesn't look like that's the problem, since the voltages around it seem OK...
Any suggestions?
Cheers
Nigel
Well, I switched the source resistors R7 to match R16 (at 0R56) and connected everything up to test... To my annoyance the DC offset was way off, and worse, the LED on the right channel started smoking....
I've been checking things out for a couple of hours, and the voltages in the right channel are all wrong - total DC offset on the output is about +18V, that is, the rail voltage. I can't believe this all happened just from changing R7....
I need a little help thinking this through, here...
Collector of Q1 is about 0.43 - which I guess has to be corrected with the trimpot, but isn't so bad, so the CCS seems OK, I guess, now I've changed the LED. So that suggests something is wrong either in the Aleph current source or the output stage around Q3. Voltages across R7 and R16 are 0.628 and 0.611, so bias current is about 1.1A, which appears to be a little higher than it was before increasing R7 from 0R47 to 0R56, which seems odd. There are no obvious shorts, and I tried switching Q8 for a new BC557 to see ifthat was the problem, but no...
Is there an easy way of checking Q3 without taking it out of the circuit? Could there be something wrong with it other than an obvious short (which the DMM doesn't show...)?
I can pull a BD139 off an old board and switch Q1 out, if neessary, but it doesn't look like that's the problem, since the voltages around it seem OK...
Any suggestions?
Cheers
Nigel
I'll use the parts designations as in sch. in post# 53. So:
Dead Q4 will smoke the LED. Dead LED means Q1 CCS is not working => LTP has very high and undefined current through it => Vgs of Q3 can be anything => Q3 can be dead => all hell broke loose
Nigel, you obviously overdid it analyzing the circuit
If above scenario is the case you'll probably have to start all over again...
Dead Q4 will smoke the LED. Dead LED means Q1 CCS is not working => LTP has very high and undefined current through it => Vgs of Q3 can be anything => Q3 can be dead => all hell broke loose
Nigel, you obviously overdid it analyzing the circuit
If above scenario is the case you'll probably have to start all over again...
Last edited:
I'll use the parts designations as in sch. in post# 53. So:
Dead Q4 will smoke the LED. Dead LED means Q1 CCS is not working => LTP has very high and undefined current through it => Vgs of Q3 can be anything => Q3 can be dead => all hell broke loose
Nigel, you obviously overdid it analyzing the circuit
You'll probably have to start all over again...
You're probably right, I messed around with things too much... but no-one ever learned very much just copying a schematic, right?
Is there an easy way to confirm that Q4 is really dead? (Or if it wasn't before, probably is now, right?
) I have enough parts to do all this again, as long as I still have a nice matched pair of BF862's. (And I'll use IRFP9140 instead of 9240 but that's OK, I guess.) Just to check, though, I only really need to replace all semiconductors, right? But I can still use the boards and passive components, so it shouldn't even take too long. I should be all right with the schematic I posted above, right? But without the 22uF, unless you tell me different, and with 9140 and BD139 in place of MJE243.... If that is all OK then it shouldn''t even be too time-consuming.
Can you confirm that that all sounds OK?
Also, the other channel (which should be OK, I hope) will have to be done to match, changin the IRFP9240s for 9140s, BD139 for MJE243, and so forth, but hopefully the matched parir in the LTP is OK. (Or is it maybe betterto jsut redo the lot?)
I also still need to understand what caused all the trouble. (I mean, apart from the obvious fact that it came from too much messing with things...
)Suppose we start from the premise that having R7 at a lower value than R16 was a bad idea. Could this (hypothetically) have caused the trouble? If not, do you have any ideas?
Cheers
Nigel
Oh, and one more query. You remarked before on the fact that I had the drain and source switched on one of the jfets. The datasheet really does claim they're interchangeable, so as long as Idss is matched remembering this, do you think I need to redo the PCBs to alter this? My guess is no, since I can't see how that could have caused the problemm but I'd like to hear what you think.
Cheers
Nigel
Cheers
Nigel
Right, that's the spirit
It's just that that you possibly learned too much Matching procedure is the way to check it
No, replace just the dead semis (Matching procedure is the way to check MOSFETs too).
Boards and passives are to be checked too.
Right, but the one from post #53 is preferable
You can leave the C2 out, it's not crucial (C2's job is to improve Vref and there's not much to improve with red LED), 9140 and BD139 are OK
Right
I see that you already understood itI also still need to understand what caused all the trouble. (I mean, apart from the obvious fact that it came from too much messing with things...
It seems to me that all the fuss was created by your attitude towards thermal behavior of JFETs. To avoid it next time, we'll try to lower their Pd:
make R1=680R and set appropriately the current through Q1 CCS (about 14mA) with R12
I personally never used that property (Source-Drain interchangeability) with BF862 and datasheets were not declared by the burning bush on Sinai.
You had a working amp with your configuration of pins, so your guess is good as mine. Anyhow, LTP is symmetrical gain block and it would be nice to use both parts in identical manner.
Well, I can check it if I can get it off the board in any condition to check... I have visions of BF862s disappearing up a soldersucker... d**n things are so da**ed small !
Oddly enough, while I don't mind matching jfets (even SMD ones) I positively hate matching mosfets... Too many d***ed wires and clips and too much heat to be accurate... If anyone has an efficient way of doing it I'd love to hear about it. Since I have 4 brand new 9140s sitting around doing nothing I'm tempted to use them and "match" the old ones another time...
Apart from manually checking the circuit, how do you "check" boards? Resistors are easy, capacitors I'll have to take on trust, since I don't have a capacitance meter. But if an electro had bitten the dust it would have made more mess, right?
I'll make the required changes - easy and cheap (we like this....
)True.
This is a little more problematic. I don't have an SMD resistor at 680R, and can't get one (at any cost-effective rate) without buying a big-ish order from Farnell-Newark in Sao Paulo. I *do* have loads of 470R and 220R, so in principle I could do 690 by putting them in series, but this wouldn't be easy (to say the least) considering they're SMD and the board is only designed for one in that spot. I can try this, but can't guarantee success
An alternative would be to use ZM's idea of the Cu U-shapes; this time I would have to take the BF862s off the boards (which I have to do on the right channel anyway) so I can then use the 120W monster iron without fear of destroying anything, and I am sure I can solder them on with that thing.
But here's the thing. ZM appears to share my concern over the jfet temperature (60 deg. C at the gate sodlering point - probably 100 deg C at the junction) but you apparently do not. Why ? Do you think this is within reasonable limits for these jfets and I am therefore worried about nothing? Obviously the easiest thing here is just to let them operate at 60 deg., ifthis is OK.
Thanks for all the help
Cheers
Nigel
Sorry, I started responding to your other post before I saw this one. I am tempted to do new boards - after jfets have lots of properties, and just because Idss is matched both ways (the ones I tested were all exactly the same if you turned them around) there is no guarantee all the other ones that might affect the sound aregoing to be equal too.
I'm in two minds about this one... I'll let you know what I do in the end.
Cheers
Nigel
Everybody would
Optically, for mechanical and electrical consistency
Good enough
I use them that way for a few months already without apparent problems. It doesn't bother me. Since it's an issue for you, go for lower Pd. 6-7mA through each JFET in LTP means considerably less dissipation and they'll still be in linear region of operation.
New PCB design
OK, so here's a PCB design adapted from the last one. The jfets now face the same way, so nouse of the symmetry of the devices. I've also decided to use a normal resistor for R1 instead of SMD, which means I can try 680R for R1 as juma suggested above. Since this gives me a jump on the board it also allows for slightly larger "heatsink" areas on all three of the jfets.
As always, all comments welcome.
juma: Why did you choose 680R as the new value for R1? It seems a rather large jump from 470R. If we have slightly larger heatsink areas on this new PCB would 560R (or something like it) be reasonable? Or still do ZMs Cu thingies instead?
Cheers
Nigel
P. S. I realise the picture has two copies of the board - sorry, it's too late to fix it now...
OK, so here's a PCB design adapted from the last one. The jfets now face the same way, so nouse of the symmetry of the devices. I've also decided to use a normal resistor for R1 instead of SMD, which means I can try 680R for R1 as juma suggested above. Since this gives me a jump on the board it also allows for slightly larger "heatsink" areas on all three of the jfets.
As always, all comments welcome.
juma: Why did you choose 680R as the new value for R1? It seems a rather large jump from 470R. If we have slightly larger heatsink areas on this new PCB would 560R (or something like it) be reasonable? Or still do ZMs Cu thingies instead?
Cheers
Nigel
P. S. I realise the picture has two copies of the board - sorry, it's too late to fix it now...
Attachments
juma: I see you replied while I was drawing the new PCB design, and I didn't look before posting. Sorry.
Are your jfets running as hot as 60 deg also? I'm probably worrying about nothing... Anyhow, the new boards will let me change the value of R1 easily, so we'll see how it goes. Cascoding the jfets is interesting, but I think I'mg going to let it wait for can wait for another project.
Cheers
Nigel
Are your jfets running as hot as 60 deg also? I'm probably worrying about nothing... Anyhow, the new boards will let me change the value of R1 easily, so we'll see how it goes. Cascoding the jfets is interesting, but I think I'mg going to let it wait for can wait for another project.
Cheers
Nigel
We want lower Id through JFET and we want to keep the same Vgs on amplifying MOSFET. The obvious way to do it is to increase R1.
Yes - a few degrees more or less (they are hot but I have no way to measure exactly)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.