Hi jackinnj,
'Iwas hoping to get your attention. Saw on your website you have made the ultra-pure oscilator in a adjustable version. Could you send me a personal email about that? Highly interested, maybe we could collaborate?
About the measurement so far:
Using a HP Vrms meter, set to 10Hz-to-100KHz bandwidth. When Input is shorted I measure 860uV rms. This should equal about 0.26nV/rtHz (G=80dB), when you take in account the first order slope beyond 100KHz (1/(pi/2) or was it 1/(sqrt(2) ?).
Using the PC sound card: sample frequency is 44100 Hz, samples used is 44100 samples, so that's a 1 second measuring time. Every measurement takes about 2 seconds in total, 10 averages to make a pretty graph so that's about 20seconds I guess. Then I measure about 0.375nV/rtHz. I'm don't know why these measurements differ. (?)
The goal is to measure THD down to -160dB using a ultra-pure oscilator, a -100dB notch filter, a low-noise amplifier and a PC sound card. Test signal will be 1KHz, and I want to measure up to 20Khz, so I guess bandwidth of interest will be 2KHz-to-20KHz
I hope I anwered you questions. (?)
Best regards,
Thijs
'Iwas hoping to get your attention. Saw on your website you have made the ultra-pure oscilator in a adjustable version. Could you send me a personal email about that? Highly interested, maybe we could collaborate?
About the measurement so far:
Using a HP Vrms meter, set to 10Hz-to-100KHz bandwidth. When Input is shorted I measure 860uV rms. This should equal about 0.26nV/rtHz (G=80dB), when you take in account the first order slope beyond 100KHz (1/(pi/2) or was it 1/(sqrt(2) ?).
Using the PC sound card: sample frequency is 44100 Hz, samples used is 44100 samples, so that's a 1 second measuring time. Every measurement takes about 2 seconds in total, 10 averages to make a pretty graph so that's about 20seconds I guess. Then I measure about 0.375nV/rtHz. I'm don't know why these measurements differ. (?)
The goal is to measure THD down to -160dB using a ultra-pure oscilator, a -100dB notch filter, a low-noise amplifier and a PC sound card. Test signal will be 1KHz, and I want to measure up to 20Khz, so I guess bandwidth of interest will be 2KHz-to-20KHz
I hope I anwered you questions. (?)
Best regards,
Thijs
I have all the pieces for the ULD-oscillator, just haven't burned the boards for it -- since I got the Boonton tweaked up I haven't needed it yet.
Nevertheless, I have a broken Tektronix SG502 which would serve as a wonderful carcase for a ULD oscillator -- I'll send you an email on the 'morrow.
Nevertheless, I have a broken Tektronix SG502 which would serve as a wonderful carcase for a ULD oscillator -- I'll send you an email on the 'morrow.
tschrama said:
Hi guys,
well after some long hours behind the PC and soldering iron I finally got to the point where I cann't find an easy increase in performance... but I am very pleased with the result, bit proud actually...🙄
I reached -144dB 2nd, 130dB 3rd and 155dB 5th harmonic distortion.. the rest is all lower...
what do you think, would this be interesting for an audioexpress article?
tschrama said:
THD meters are more useful than any ears or simulations (my opinion and statement)
for audio project evaluations.
VERY Good project idea and effort, tschrama!
peranders said:
peranders ask a valid question here.
But, it is plain obvious answer:
we want, in diyaudio, know noise figures and THD figures
in manily area for what we can HEAR, and some factors above this.
Therefore we call it audio - from human point of view.
Other animals can define 'audio' very differently.
runebivrin said:
If 'there is no reason believe it wont work for audio'
then all is very well.
tschrama said:
Dreams, visions and what you want are
the beginning of most progress done by humans.
JFETs are not mainly known as used for low noise.
But new better JFETs comes all the time!
Parallelling technique is good and will lower noise,
just as much when used with bipolar, BJT.
amplifierguru said:
Greg can not think of anything else than 'his perfect future amp'. 🙂
Even if here we talk about a THD meter project in first place.
But he is right, in that also such a meter has to amplify, considerably!
bocka said:
And if I parallell 8 x BC550C, I will get .... ??
tschrama said:
tschrama has thought a bit further.
Right he is!
If it wasnt for input capacitance figures and a few other drawbacks
MOSFETs and JFETs would rule the world
and most none would longer use bipolar transistors no more.
tschrama said:
A probably realistic point when doing FET matching.
If you can use 1/5 JFET to get a good match
and use 1/2 BC550C bipolar to get close enough match.
Then means you should multiply price tag of 100 JFETs devices
with factor x 2.5
when you compare to BC550C / BC560C
BEFORE buying your 100-pack.
----------------------------
If you havent read Rod Elliott basic simple THD meter project page
then you all should, if you are interested in distortion measuring.
He says his already very basic circuit 'can be simplified'.
Distortion Analyser- ESP Project 52
thats it, from me, my $500 USA-dollars thoughts
🙂
Hi Lineup.
thanks for ll your info! Actually it was Rod elliots Project that gor me started. The Notch filter I use is actually a scaled copy of his project.
After some experimenting I have reached -141dB for the highest harmonic using a passive lowpass filter at the output. DynRange is about 160dB using 10sec intergartion time. Mostly limited by the noise of the 10K resistors in the notched filter.
The ultra-low-noise amplifier is finished but the input capacitance has been troublesome in conjuction with the passive notch filter, so the next step will be to use two opa227s for a 80dB low noise, low THD gain.
BTW I also used the IF3601 lownoise JFET for an low noise amplifier and it works like a jam...
The opa227 has proved it self to be supperior to the opa627 THD and noise wise in my oscilator. That, combined with th lower currect draw and higher swing to the voltage supply rails, makes it my favorite opamp of all.
Next up is a comparison with the HP334a as a bench mark.. maybe tweking the 334a with modern components.
I would really like to see I audioXpress would be interested, but to be honest, I haven't done anything original. I have gotten these specs by keeping the circuitery as simple as possible and optimizing the limmiting ellements.
regards,
Thijs
thanks for ll your info! Actually it was Rod elliots Project that gor me started. The Notch filter I use is actually a scaled copy of his project.
After some experimenting I have reached -141dB for the highest harmonic using a passive lowpass filter at the output. DynRange is about 160dB using 10sec intergartion time. Mostly limited by the noise of the 10K resistors in the notched filter.
The ultra-low-noise amplifier is finished but the input capacitance has been troublesome in conjuction with the passive notch filter, so the next step will be to use two opa227s for a 80dB low noise, low THD gain.
BTW I also used the IF3601 lownoise JFET for an low noise amplifier and it works like a jam...
The opa227 has proved it self to be supperior to the opa627 THD and noise wise in my oscilator. That, combined with th lower currect draw and higher swing to the voltage supply rails, makes it my favorite opamp of all.
Next up is a comparison with the HP334a as a bench mark.. maybe tweking the 334a with modern components.
I would really like to see I audioXpress would be interested, but to be honest, I haven't done anything original. I have gotten these specs by keeping the circuitery as simple as possible and optimizing the limmiting ellements.
regards,
Thijs
Did you use some sort of common-mode cancellation like presented in linears appnotes?tschrama said:
On my oscillator OPA228 didnt have any change against OPA637, OPA637 was something like 15db better on 2nd harmonic.
That's interesting!
I used the opa227 with 1Vrms output, gain was 3x, power supply was from two 9Volt batteries. Load was about 1KOhm... I did notice that when the RMS output voltage is alowed to ris beyong 1.5Vrms, distortion suddenly rises quickly.
This might caused by the output transistor of the opa227 going into classB, while staying in Class A when they circuit is limmited to less than 1Vrms output.
I did notice that the OPA627 has better current output capabilities, suggesting higher Iq setting of the output transistors.
THD was purely 3rd Harmonic with the opa627, which was a suprrise to me at first. The opa227 has some second harmonic too. I tried the opa637, but as expected it was not stable at a gain of 3x, and HF oscilation occured.
I tried the dual-imverted type wienbridge also, but I couldn't get distortion lower then -110dB. the classic non-inverting wienbridge was cleary superior, using the same components. Why.... i don't know....
How did you get the opa637 working low gain, or didn't you use a classic wienbridge?
What power supply , output load and voltage swing, where you using?
If all is optimized, the third harmonic should be the highest harmonic.
,grzz,
Thijs
edit: my mistake I I said opa227 anywhere.. I meant opa228. I proved to be stable at gain of 3x.
I used the opa227 with 1Vrms output, gain was 3x, power supply was from two 9Volt batteries. Load was about 1KOhm... I did notice that when the RMS output voltage is alowed to ris beyong 1.5Vrms, distortion suddenly rises quickly.
This might caused by the output transistor of the opa227 going into classB, while staying in Class A when they circuit is limmited to less than 1Vrms output.
I did notice that the OPA627 has better current output capabilities, suggesting higher Iq setting of the output transistors.
THD was purely 3rd Harmonic with the opa627, which was a suprrise to me at first. The opa227 has some second harmonic too. I tried the opa637, but as expected it was not stable at a gain of 3x, and HF oscilation occured.
I tried the dual-imverted type wienbridge also, but I couldn't get distortion lower then -110dB. the classic non-inverting wienbridge was cleary superior, using the same components. Why.... i don't know....
How did you get the opa637 working low gain, or didn't you use a classic wienbridge?
What power supply , output load and voltage swing, where you using?
If all is optimized, the third harmonic should be the highest harmonic.
,grzz,
Thijs
edit: my mistake I I said opa227 anywhere.. I meant opa228. I proved to be stable at gain of 3x.
hey.
Interesting to know about your OPA227 findings, tschrama!
If it works so well,
I would also consider similiar biploar OP-amps
OP37, OPA37 which I have downloaded PDF datasheets of.
I always go for the low-to-medium priced lines
(normal diyaudio people and audio interested students are not made out of money, you know)
so I may sometimes miss the advanced edge technology chips.
When it comes to low noise high gain op-amps
we have the proven LT1028, LT1115 op-amps
(wellknown by peranders - he actually is not very bad! when come to low noise designs and pc-boards)
http://home.swipnet.se/~w-50719/hifi/index.html
Basically in this we should go for same devices as in good RIAA Phono amplifiers.
You should NOT forget, the practical PC board or implementations
in these very low noise levels
are also very important for final result.
Screening and good grounding are vital concerns.
finally what Nelson Pass
The one And Only
has to say on good signal generator (1 kHz oscillators)
Regards to you tschrama,
whoever you are and wherever your loation is
from
lineup, at home right now
😀
Interesting to know about your OPA227 findings, tschrama!
If it works so well,
I would also consider similiar biploar OP-amps
OP37, OPA37 which I have downloaded PDF datasheets of.
I always go for the low-to-medium priced lines
(normal diyaudio people and audio interested students are not made out of money, you know)
so I may sometimes miss the advanced edge technology chips.
When it comes to low noise high gain op-amps
we have the proven LT1028, LT1115 op-amps
(wellknown by peranders - he actually is not very bad! when come to low noise designs and pc-boards)
http://home.swipnet.se/~w-50719/hifi/index.html
Basically in this we should go for same devices as in good RIAA Phono amplifiers.
You should NOT forget, the practical PC board or implementations
in these very low noise levels
are also very important for final result.
Screening and good grounding are vital concerns.
finally what Nelson Pass
The one And Only
has to say on good signal generator (1 kHz oscillators)
http://www.diyaudio.com/forums/showthread.php?postid=83985#post83985Nelson Pass said:
Regards to you tschrama,
whoever you are and wherever your loation is
from
lineup, at home right now
😀
here is another postive post on
OPA227
in this forum
http://www.diyaudio.com/forums/showthread.php?postid=15804#post15804
cheers
lineup
OPA227
in this forum
http://www.diyaudio.com/forums/showthread.php?postid=15804#post15804
cheers
lineup
I used modified wien-bridge with gain of 5.tschrama said:
2x9v batteries, load was about 1.5k (passive notch.)
Output voltage about 1Vrms.
With OPA637 only harmonic over my -150db measurement limit was 2nd harmonic, apprx at -138db
On my tests MAX437 was outperforming OPA228 clearly. OP37 from AD was complete disaster in performance, while maxim's OP37 performed decades better than AD's. (AD's OP37 was possibly quite old)
I havent tested yet with +-15 supplies, there could be a small improvement even with low output level like 1vrms...
Edit: I have been also planning to try buffered version with BUF634 or my discrete schizklai(spelling?) buffer.
Hi,
those are all high quality opamps. I made an effort to try all of my opamps that I have in stock like the ad797, op37, opa134, lt1115, lt1028 (oscilated), opa627, NE5534, (not bad at all, 3rd place), 741
, .. but opa228 was the winner, allthouhg opa627 was allmost just as good..
The only reason I can think of that my opa228 is performing so good, is that current load for the opamp is low, so the opamp stays in class A.
Grz,
Thijs..sitting in his office reading the HP334 service manual, simulating the impedance buffer in SPICE 🙂
those are all high quality opamps. I made an effort to try all of my opamps that I have in stock like the ad797, op37, opa134, lt1115, lt1028 (oscilated), opa627, NE5534, (not bad at all, 3rd place), 741
, .. but opa228 was the winner, allthouhg opa627 was allmost just as good..The only reason I can think of that my opa228 is performing so good, is that current load for the opamp is low, so the opamp stays in class A.
Grz,
Thijs..sitting in his office reading the HP334 service manual, simulating the impedance buffer in SPICE 🙂
tschrama said:
Well, thats easy. Just scale parallei rc-combination from non-inverting input to gnd to be 2C and 0.5R compared to orginal wien-bridge. I am not sure if it should be still called wien-bridge or just rc-oscillator tough..
I am a bit lazy to write emails so put this up here if others are also intersted. 😉
Thijs, I forgot to ask what kind of like amplitude stabilation system you use? LDR-photocell or some kinda AGC-amplifier?
Thx, I would like to try that and compare the results. I use the
classic lamp bulb ... allthough I would be interested in using a simple but good the LDR circuit abd see if that works better. do you use LDR?
grzz,
Thijs
classic lamp bulb ... allthough I would be interested in using a simple but good the LDR circuit abd see if that works better. do you use LDR?
grzz,
Thijs
Wow, those are impressive results from bulb-stabilation. I tried it but it was useless at best. If I trimmed it for minimum distortion it didnt start up. Or amplitude stabilation took a year to settle. And it was VERY microphonic, smallest knock make it stop oscillation or amplitude got trough the roof. You probably have much better bulb than I have, I tried plenty of different low-current signal light bulbs but noone were satisfying. I woulnt use bulb-stabilation for anything more serious...tschrama said:
Yeap, I use LDR. Gives you much more control over stability and amplitude settling as you can tweak PI-controllers parameters just by swapping resistors and caps. LDR's are far away form linear resistors so you want to keep voltage over them at minimum. So LDR's control range needs to be real small, if you have 2k resistor on feedback your LDR should have something like +-1% or less adjustement range to that 2k resistor.
Silonex http://www1.silonex.com/audiohm/index.html
Have potential looking LDR's ,better than ones I have been using(I was lazy to order those silonex´s from Farnell)
yeh, I have had all the same problems and it took me a while to figure that out. .. but I got around to that with carefull component selection, send me a pm...
Judging from the appnote, even with those LDRs it will be difficult to got down to below -140dB.. or maybe, but I have yet to experiment with that...
grzz,
T
Judging from the appnote, even with those LDRs it will be difficult to got down to below -140dB.. or maybe, but I have yet to experiment with that...
grzz,
T
tschrama said:
Below -140db is nothing but trouble, opamps are running out of feedback factor, cap selection becomes critical, even some resistors can cause distortion...
Linears ultra-pure oscillator with 180db of feedback is the way to go below -140db. I am affraid that cap selection becomes quite tricky for 1khz design if you take this road. Good thing would be that you could bump up the feedback factor to insane 210db on 1khz desing. Only limiting factor is going to be passive component's linearity/distortion, opamps are not problem anymore.
mzzj said:
The Linear Ultra Low Distortion Oscillator uses an LT1228 CFB adjustable gain amplifier for amplitude control -- these are $6.88 in single quantities from Digikey. Cyril Bateman's LD oscillator from his "Capacitor Sound" article in EW uses AD797's (or NE5532's if you don't want to pay up) and the Analog Devices SSM2018 VCA which is $4.45 from DK.
Anyone ever made a GIC oscillate? There was a GIC based oscillator in EDN (Lutz von Wangenheim from University of Applied Sciences in Bremen) but I can't get it to work, and I can't get it to simulate either. I understand the principle of GIC's and FDNR's but can't seem to get this one correct.
Never before heard of a GIC oscilator... I did found the EDN version... but I sorry I cann't say I fully understand it.
- Status
- Not open for further replies.