we are agreed, your warm output offset is ~-175mVdc.
What is it cold?
What is it with the servo working?
What is it cold?
What is it with the servo working?
Yes, after over 30 minutes of warmup and stabilised temperature, the offset was 175mV.
it was exactly the same cold, and so was the voltage across the emitter resistors, nothing changed.
When I, initially, had the servo hooked up and no Wahab-mods to the amplifier, the offset would surge to around -7V on power-up, before settling down to about 1,2 V after some 10-15 seconds.
it was exactly the same cold, and so was the voltage across the emitter resistors, nothing changed.
When I, initially, had the servo hooked up and no Wahab-mods to the amplifier, the offset would surge to around -7V on power-up, before settling down to about 1,2 V after some 10-15 seconds.
So far, i simulated the amp..
As already pointed, the output DC offset is the by product
of the negative feedback DC loop trying to cancel the input differentials
transistors mismatch..
To obtain such a high offset, T1 must be badly mismatched with T2
while T3 must have as much mismatch , but with T4 of course..
The problem lie on matching the input pairs, T1 with T2 and T3 with T4.
Matching of T1/T2 in respect of T3/T4 is less important but is worthy if
there s enough components available for matchings..
As already pointed, the output DC offset is the by product
of the negative feedback DC loop trying to cancel the input differentials
transistors mismatch..
To obtain such a high offset, T1 must be badly mismatched with T2
while T3 must have as much mismatch , but with T4 of course..
The problem lie on matching the input pairs, T1 with T2 and T3 with T4.
Matching of T1/T2 in respect of T3/T4 is less important but is worthy if
there s enough components available for matchings..
Wahab,
I have no reason to doubt your conclusion here, allthough I am a little bit surprised that this amp requires matched transistors after all.. or perhaps the transistors I bought were of really poor quality??
Now, transistor matching is not something I've done too much of.. what parameters should be matched and what would be acceptable tolerances/ ranges??
I have no reason to doubt your conclusion here, allthough I am a little bit surprised that this amp requires matched transistors after all.. or perhaps the transistors I bought were of really poor quality??
Now, transistor matching is not something I've done too much of.. what parameters should be matched and what would be acceptable tolerances/ ranges??
This kind of amp need matched differential pairs , even if it use
a servo , as a balanced differential has minimal distorsion.
The DC servo is used only to get rid of the cap that you
put in serial with R22.
Now, it s impossible to tell you if your bunch of components
are a bad serie..
If you unsold the four transistors, please report their beta
measured with your tester.
You can use your beta tester to select a first set of transistors.
Take the ones with higher and closer beta at the same time.
Uh.. what is "beta"? 😱
It makes perfect sense to remove these transistors and measure them, any "bad" values that could explain the offset problem would be the closing evidence for this case I guess..
It makes perfect sense to remove these transistors and measure them, any "bad" values that could explain the offset problem would be the closing evidence for this case I guess..
Beta is the current gain , seems there s a beta tester in you MM.
Attachments
Bad match of beta will show as a different Vdrop across the base resistors.
Bad match of Vbe will show as Vdiff between +IN & -IN.
A feedback amplifier tries to set the output voltage to minimise the Vdiff at the inputs. A Vbe difference in the two input devices will result in an output offset.
Different currents through the +IN transistor from the -IN transistor will make the matching worthless.
Bad match of Vbe will show as Vdiff between +IN & -IN.
A feedback amplifier tries to set the output voltage to minimise the Vdiff at the inputs. A Vbe difference in the two input devices will result in an output offset.
Different currents through the +IN transistor from the -IN transistor will make the matching worthless.
Ah, hFE... then I get it.... 🙂 But what sort of spread/ range should I aim for??
I quickly checked 18 BC546 b transistors I had left over (T1-2), and they varied from 322 to 380..
I quickly checked 18 BC546 b transistors I had left over (T1-2), and they varied from 322 to 380..
Last edited:
get operating Ic within 5% when both Vbe of the LTP are identical.
Get hFE within 10%.
I'd start by grouping the hFE using the DMM.
Then test/measure hFE at the operating current that the circuit uses.
Then pick out groups of hFE that are within 10% or better at operating current and set a pair up as DUT + REF in an LTP configuration with emitters connected, with bases connected, with a pair of 0.1% 1k0 as load resistors, one to each collector.
Now adjust Vbe until Ic is correct and measure the Vdiff at the collectors.
If you have a very good match, you can adjust Vbe to swing a range of Ic and the collector Vdiff will hardly change.
Get hFE within 10%.
I'd start by grouping the hFE using the DMM.
Then test/measure hFE at the operating current that the circuit uses.
Then pick out groups of hFE that are within 10% or better at operating current and set a pair up as DUT + REF in an LTP configuration with emitters connected, with bases connected, with a pair of 0.1% 1k0 as load resistors, one to each collector.
Now adjust Vbe until Ic is correct and measure the Vdiff at the collectors.
If you have a very good match, you can adjust Vbe to swing a range of Ic and the collector Vdiff will hardly change.
Looks like I have a bit of a job ahead of me then.. I've got six of these amplifier cards..
Sorry for being stupid again, but what is LTP?😱
Sorry for being stupid again, but what is LTP?😱
My mistake.
Long Tail Pair (LTP) is the way to use first introduction to an acronym and thereafter the acronym can be used to save typing.
Long Tail Pair (LTP) is the way to use first introduction to an acronym and thereafter the acronym can be used to save typing.
No problem Andrew, I realize you are at quite a different level than I am! 🙂
My next question is of course, what is a long tail pair?😕
My next question is of course, what is a long tail pair?😕
So the zipped schematic describes what a LTP looks like..
While I ponder that one for a bit, does anybody know if it is possible to buy matched transistors anywhere ?
While I ponder that one for a bit, does anybody know if it is possible to buy matched transistors anywhere ?
You ll surely have to DIY....
This is not very hard since it seems that the ones on
your amp are fairly unmatched as they were unselected,
so you can only improve the situation..
After all, that s an occasion to gain some experience
in the matter..
I agree, looks like this is something I need to learn...
But for starters, I'll remove T1-4 from the card and measure them
But for starters, I'll remove T1-4 from the card and measure them
Well, out went T1-4, and i just used the multimeter to do a quick check on them just to see if this would reveal something obvious.
T1 = Vbe 0,943 Hfe 357
T2 = Vbe 0,939 Hfe 308
T3 = Vbe 0,998 Hfe 255
T4 = Vbe 0,993 Hfe 256
Now, as I mentioned earlier on, the Vbe values seem a bit high, but I'll put that down to the multimeter and assume that this is usable for relative comparison.
What is obvious is that there was quite a difference in Hfe between T1 and T2
T3 and T4 seemed to be a better match, but the difference in Hfe Between the T1/2 pair and the T3/4 pair seems quite big.
But the big question is, and I need help here, does this correlate with and fully explain the -0,175 V offset on the output?
T1 = Vbe 0,943 Hfe 357
T2 = Vbe 0,939 Hfe 308
T3 = Vbe 0,998 Hfe 255
T4 = Vbe 0,993 Hfe 256
Now, as I mentioned earlier on, the Vbe values seem a bit high, but I'll put that down to the multimeter and assume that this is usable for relative comparison.
What is obvious is that there was quite a difference in Hfe between T1 and T2
T3 and T4 seemed to be a better match, but the difference in Hfe Between the T1/2 pair and the T3/4 pair seems quite big.
But the big question is, and I need help here, does this correlate with and fully explain the -0,175 V offset on the output?
Last edited:
You should simply first change the battery of your multimeter..
The readed value can be stuck at a high value whatever the real
vbe of the tested device...
Finding the solution using a faulty tester is just plain illogism,
moreover if the devices must be matched using this same apparatus.
The readed value can be stuck at a high value whatever the real
vbe of the tested device...
Finding the solution using a faulty tester is just plain illogism,
moreover if the devices must be matched using this same apparatus.
- Status
- Not open for further replies.