Svart
Well-known member
glad to see you around again TK. I've been steadily using up my FET blokes on just about everything these days.. almost time for a refill!
Another Discrete Amp - GainBloak
- Thread starter tk@halmi
- Start date
Help Support GroupDIY Audio Forum:
Samuel Groner
Well-known member
Hm, is there really a serious stability difference? There should only be a minor difference, as with this topology stability/loop gain is essentially given by feedback resistor value, collector junction capacity and perhaps output loading. Transconductance of the input transistor (and hence BJT vs. FET) is almost irrelevant, at least at low gains where stability is of most concern.
Samuel
[quote author="Samuel Groner"]
Samuel[/quote]
While perhaps not stability per se that kind of agrees with my recollection of the basic difference between FET and Bipolar inputs. i.e. the lower transconductance of FETs allow higher slew rate under similar conditions due to reduced dominant pole compensation requirements. This lower transconductance can be accomplished in bipolar input designs by adding emitter degeneration resistors as was done in old LM318 (1k emitter degeneration on each input device). This 50+V/usec speed of the LM318 didn't come without a price as the series R increased offest voltage and input noise.
Of course this is based on my now shaky understanding of stability criteria. :roll:
JR
Hm, is there really a serious stability difference? There should only be a minor difference, as with this topology stability/loop gain is essentially given by feedback resistor value, collector junction capacity and perhaps output loading. Transconductance of the input transistor (and hence BJT vs. FET) is almost irrelevant, at least at low gains where stability is of most concern.
Samuel[/quote]
While perhaps not stability per se that kind of agrees with my recollection of the basic difference between FET and Bipolar inputs. i.e. the lower transconductance of FETs allow higher slew rate under similar conditions due to reduced dominant pole compensation requirements. This lower transconductance can be accomplished in bipolar input designs by adding emitter degeneration resistors as was done in old LM318 (1k emitter degeneration on each input device). This 50+V/usec speed of the LM318 didn't come without a price as the series R increased offest voltage and input noise.
Of course this is based on my now shaky understanding of stability criteria. :roll:
JR
Samuel Groner
Well-known member
For differential inputs yes, but here we have a single-ended current-feedback input.
Samuel
hmmm, getting lost overhere...
after several attempts i made no progress trying to build line preamp using fet input (k170BL not 2sk170bl- but they should be similar?) - darlington output (mje 700-800) and surrounding circuit (below)...
Everything is working, it passes audio, no hum or other problems but
when i tried to get some measurement, this is what i ended up with:
What am i doing wrong? The distortion is very hi as you could see...
Using +/-24vdc PSU, no excessive heat builds up on any of the components. I tried to add series resistance to cable (180r) like Bcarso recommended, but RMAA responded with even worse results...
Any help greatly appreciated.
Thank you
after several attempts i made no progress trying to build line preamp using fet input (k170BL not 2sk170bl- but they should be similar?) - darlington output (mje 700-800) and surrounding circuit (below)...
Everything is working, it passes audio, no hum or other problems but
when i tried to get some measurement, this is what i ended up with:
What am i doing wrong? The distortion is very hi as you could see...
Using +/-24vdc PSU, no excessive heat builds up on any of the components. I tried to add series resistance to cable (180r) like Bcarso recommended, but RMAA responded with even worse results...
Any help greatly appreciated.
Thank you
tk@halmi
Well-known member
Are you using four (4) diodes to bias the darlingtons? If you use only two the output is not biased corerctly.
Chevapci!
tamas
Chevapci!
tamas
Samuel Groner
Well-known member
Without specifying gain, level, frequency and measurement bandwidth a THD+N figure is completely pointless.
Samuel
Samuel
tk@halmi
Well-known member
If syn is using the diagram shown the max gain is (R1/R2)+1, about 11, not a problem for any of the Bloaks to give 0.01% distortion or less. You would have to be in the 60dB range to get a significant rise in distortion using the darlington output version.
By default RMAA measures distortion at 1Khz although one can change it to go up to 7KHz, if I recall right.
I noticed that you are using 5k feedback network. This increases the bandwidth of the opamp greatly. I have run them as low as 2.5k, but you need a really neat layout and awsome decupling to keep stability. I usually run them 10k or larger just to make life easier.
Try reducing the supply rails to +-20V and test again. Let me know what happens.
By default RMAA measures distortion at 1Khz although one can change it to go up to 7KHz, if I recall right.
I noticed that you are using 5k feedback network. This increases the bandwidth of the opamp greatly. I have run them as low as 2.5k, but you need a really neat layout and awsome decupling to keep stability. I usually run them 10k or larger just to make life easier.
Try reducing the supply rails to +-20V and test again. Let me know what happens.
Samuel Groner
Well-known member
Instability was my first ideas as well but as he mentioned everything is working, it passes audio, no hum or other problems this seems to be ruled out. I suspect that the sweep is simply clipped somewhere, that's why I asked about levels.
Samuel
Samuel
Thanks for the replays.
Samuel no clipping has occurred. RMAA 6.0.5 informs you if that happens and stops the test until you press OK button...
I made a mistake of reversing all 4 bias diodes
, that was the reason of yesterdays THD disaster...
After testing in various configurations all day today, distortion is still high,
and whatever i do does not bring it down...tried with transformer and without it, tried 679a+680a, 139+140 (even worse than mje's).
The pre is constructed after the above diagram...
No heat builds anywhere
c1 4700uF/35V nichicon PW series ( jamicon 1000uF/63V)
c2 1500uF/35V nichicon PW series ( jamicon 1000uF/63V)
20 Hz - 20 kHz filter OFF
Normalize amplitude ON
Mono mode ON
Calibration signal, Hz 1000
Polarity correct/correct
input level ~ -1.5dB
+/-20VDC PSU
stancor 4:1 on input (2k7 load resistor)
k170
mje700/800 out
R1 10k
r2 470r
SummaryFrequency response (from 40 Hz to 15 kHz), dB +0.03, -0.13 Very good
From 20 Hz to 20 kHz, dB -0.37, +0.10
Noise level, dB (A) -91.7 Very good
Dynamic range, dB (A) 90.4 Very good
THD, % 0.387 Poor
THD + Noise, dB (A) -45.9 Poor
IMD + Noise, % 0.421 Poor
IMD at 10 kHz, % 0.348 Average
General performance Good
+/-24VDC PSU
stancor 4:1 on input(2k7 load resistor)
k170
mje700/800 out
r1 10k
r2 1k
Frequency response (from 40 Hz to 15 kHz), dB +0.03, -0.10 Excellent
Noise level, dB (A) -89.5 Good
Dynamic range, dB (A) 89.1 Good
THD, % 0.357 Poor
THD + Noise, dB (A) -46.6 Poor
IMD + Noise, % 0.307 Average
IMD at 10 kHz, % 0.323 Average
General performance Good
+/-24VDC PSU
k170 on input (2.7k "in" resistor)
mje700/800 out
r1 5k
r2 470r
Frequency response (from 40 Hz to 15 kHz), dB +0.02, -0.09 Excellent
Noise level, dB (A) -96.7 Excellent
Dynamic range, dB (A) 95.0 Very good
THD, % 0.186 Average
THD + Noise, dB (A) -52.3 Poor
IMD + Noise, % 0.167 Average
IMD at 10 kHz, % 0.170 Average
General performance Very good
These are some samples of tests i caried...
I'm still wrong somewhere... Is it possible that my layout sucks(above)
that much?
tk@halmi Chevapci!? You like it? If you ever come over here let me know,
i'll take you to the chevapci heaven...
Thank you
Samuel no clipping has occurred. RMAA 6.0.5 informs you if that happens and stops the test until you press OK button...
I made a mistake of reversing all 4 bias diodes
, that was the reason of yesterdays THD disaster...After testing in various configurations all day today, distortion is still high,
and whatever i do does not bring it down...tried with transformer and without it, tried 679a+680a, 139+140 (even worse than mje's).
The pre is constructed after the above diagram...
No heat builds anywhere
c1 4700uF/35V nichicon PW series ( jamicon 1000uF/63V)
c2 1500uF/35V nichicon PW series ( jamicon 1000uF/63V)
20 Hz - 20 kHz filter OFF
Normalize amplitude ON
Mono mode ON
Calibration signal, Hz 1000
Polarity correct/correct
input level ~ -1.5dB
+/-20VDC PSU
stancor 4:1 on input (2k7 load resistor)
k170
mje700/800 out
R1 10k
r2 470r
SummaryFrequency response (from 40 Hz to 15 kHz), dB +0.03, -0.13 Very good
From 20 Hz to 20 kHz, dB -0.37, +0.10
Noise level, dB (A) -91.7 Very good
Dynamic range, dB (A) 90.4 Very good
THD, % 0.387 Poor
THD + Noise, dB (A) -45.9 Poor
IMD + Noise, % 0.421 Poor
IMD at 10 kHz, % 0.348 Average
General performance Good
+/-24VDC PSU
stancor 4:1 on input(2k7 load resistor)
k170
mje700/800 out
r1 10k
r2 1k
Frequency response (from 40 Hz to 15 kHz), dB +0.03, -0.10 Excellent
Noise level, dB (A) -89.5 Good
Dynamic range, dB (A) 89.1 Good
THD, % 0.357 Poor
THD + Noise, dB (A) -46.6 Poor
IMD + Noise, % 0.307 Average
IMD at 10 kHz, % 0.323 Average
General performance Good
+/-24VDC PSU
k170 on input (2.7k "in" resistor)
mje700/800 out
r1 5k
r2 470r
Frequency response (from 40 Hz to 15 kHz), dB +0.02, -0.09 Excellent
Noise level, dB (A) -96.7 Excellent
Dynamic range, dB (A) 95.0 Very good
THD, % 0.186 Average
THD + Noise, dB (A) -52.3 Poor
IMD + Noise, % 0.167 Average
IMD at 10 kHz, % 0.170 Average
General performance Very good
These are some samples of tests i caried...
I'm still wrong somewhere... Is it possible that my layout sucks(above)
that much?
tk@halmi Chevapci!? You like it? If you ever come over here let me know,
i'll take you to the chevapci heaven...
Thank you
Samuel Groner
Well-known member
I still miss the info on level, gain and bandwidth (or did I overread that?). This info is really important and the specs don't make much sense without it (yeah, one should tell that the Rightmark people).
Do you really have the 600 ohm load at the output as shown in the schematic?
A few things you might check and report:
* 100 kHz square wave response
* quiescent current of the opamp
* THD+N versus frequency
* THD+N versus level at 100 Hz, 1 kHz and 10 kHz (if the measurement bandwidth is at least 40 kHz)
* spectrum of the distortion at 1 kHz
The most helpful thing would be to look at the residuum, but i guess that's not possible with the software.
Samuel
Do you really have the 600 ohm load at the output as shown in the schematic?
A few things you might check and report:
* 100 kHz square wave response
* quiescent current of the opamp
* THD+N versus frequency
* THD+N versus level at 100 Hz, 1 kHz and 10 kHz (if the measurement bandwidth is at least 40 kHz)
* spectrum of the distortion at 1 kHz
The most helpful thing would be to look at the residuum, but i guess that's not possible with the software.
Samuel
Samuel
I'll try to answer:
Input level to sound card and RMAA was ~1.5dB
yes i have 610ohm resistor on the output (out to ground) as in the schematic.
gain1 was : ~20
gain2 was : ~10
gain3 was : ~10
bandwith around 44-96kHz (sound card was in 88-192kHz mode)...
100kHz square wave response - guess I need scope for that?
quiscent current of the op amp - how to masure it?
THD+N versus frequency
THD+N versus level at 100 Hz, 1 kHz and 10 kHz (if the measurement bandwidth is at least 40 kHz)
spectrum of the distortion at 1 kHz
guess those could be seen in the RMAA reports? Download the file below please.
RMAAreports
Thank you
I'll try to answer:
Input level to sound card and RMAA was ~1.5dB
yes i have 610ohm resistor on the output (out to ground) as in the schematic.
gain1 was : ~20
gain2 was : ~10
gain3 was : ~10
bandwith around 44-96kHz (sound card was in 88-192kHz mode)...
100kHz square wave response - guess I need scope for that?
quiscent current of the op amp - how to masure it?
THD+N versus frequency
THD+N versus level at 100 Hz, 1 kHz and 10 kHz (if the measurement bandwidth is at least 40 kHz)
spectrum of the distortion at 1 kHz
guess those could be seen in the RMAA reports? Download the file below please.
RMAAreports
Thank you
skipwave
Well-known member
[quote author="tk@halmi"]Are you using four (4) diodes to bias the darlingtons? If you use only two the output is not biased corerctly.[/quote]
Is this not the issue?
Is this not the issue?
Samuel Groner
Well-known member
11 mA is pretty low. If you have the 600 ohm load as shown in the schematic there might be sufficient output stage distortion to cause these numbers. A quick look (will try to find more time for it but not possible now) at your measurement data suggest that there is som hum as well which will add to the figures.
Samuel
Samuel
tk@halmi
Well-known member
Hi Syn,
Samuel is right. The quiescent current should be at least 30ma based on measurements. The symptoms seem similar to not having much current run through the input FET, throwing off the biasing for the entire amp. Are you sure that 1k is not 10k or something nutty like that?
Can you measure voltage on the source of the 170BL and the emitter of the BC560 transistors?
Tamas
Samuel is right. The quiescent current should be at least 30ma based on measurements. The symptoms seem similar to not having much current run through the input FET, throwing off the biasing for the entire amp. Are you sure that 1k is not 10k or something nutty like that?
Can you measure voltage on the source of the 170BL and the emitter of the BC560 transistors?
Tamas
Thank you very much, guys
"1k" is 1k.
K170BL source 0.316vdc
BC560 emitter +22.4 vdc
another thing, if i remove 600ohm load, the specs get much much better...
maudio card only output to input
THD, % 0.011 Good
THD + Noise, dB (A) -77.6 Average
IMD + Noise, % 0.016 Very good
IMD at 10 kHz, % 0.015 Very good
bloak no 600ohm load on the output
THD, % 0.015 Good
THD + Noise, dB (A) -74.5 Average
IMD + Noise, % 0.020 Good
IMD at 10 kHz, % 0.018 Very good
"1k" is 1k.
K170BL source 0.316vdc
BC560 emitter +22.4 vdc
another thing, if i remove 600ohm load, the specs get much much better...
maudio card only output to input
THD, % 0.011 Good
THD + Noise, dB (A) -77.6 Average
IMD + Noise, % 0.016 Very good
IMD at 10 kHz, % 0.015 Very good
bloak no 600ohm load on the output
THD, % 0.015 Good
THD + Noise, dB (A) -74.5 Average
IMD + Noise, % 0.020 Good
IMD at 10 kHz, % 0.018 Very good
solved! somehow i mirrored the output stage . Anyhow the Bloak measures now as good as my sound card, like there is a wire only...
Transients are back like i remember them when testing it on a protoboard some time ago...
Over 50mA of quiescent current (i have 19k22 instead of d7)
after few minutes drops to 36-37mA.
One last question how many mA of quiescent current is max (+/-24vdc)?
Tamas and Samuel thank you very much on your kind help.
:guinness: :sam: :guinness: :sam: :guinness: :sam: :guinness: :sam:
Transients are back like i remember them when testing it on a protoboard some time ago...
Over 50mA of quiescent current (i have 19k22 instead of d7)
after few minutes drops to 36-37mA.
One last question how many mA of quiescent current is max (+/-24vdc)?
Tamas and Samuel thank you very much on your kind help.
:guinness: :sam: :guinness: :sam: :guinness: :sam: :guinness: :sam:
