You are a gem for the hardcore electronics enthusiastic... your new videos always makes me smiling...please make a video on how you achieve all these high skilled stuffs... Thank you so much Sir...
Great video, great explanations as always! I'm pleasantly surprised how much this makes sense given that I took the relevant college classes 40+ years ago and have worked exclusively in the digital domain since then.
Really cool video, and a creative way to measure the THD. I am not as familiar with the lock in amplifiers. I am just wondering if it is still possible that the THD may be worse on other ranges and frequencies of the DS360. Perhaps the signal path used for the 1.4V 775hz does not go through the resistors which reached high temperatures?
Amazing piece of kit, nice troubleshooting and an easy problem to repair. The biggest flex in this video in my opinion is the little "Bench 8" sticker on the table, I wish I had as much benchspace. ;)
A very good video as per. I really like the SRS equipment UIs and industrial design. The notch method of THD measurement seems so obvious... once it's explained. Thank you for doing so.
3458A is also very capable to measure THD, with floor below -110dB at 1kHz or there about. There are some examples available with GPIB code for this application.
Cool video. I'm not familiar with lock-in amps, so I'll have to watch that other video but using a notch and adding gain seems like an intuitive solution to get the measure. I haven't worked with a lot of analog equipment, but I have written phase accumulators and worked with SIMD math libraries for audio synthesis before. Cool to see a practical DDS application for lab use.
So glad the repair wasn't just another power supply, but alas just an opto . . . You gotta get that rental place to send you another one that they can't fix. But still enjoyable to see how you verified it.
Really nice video. The main contributor to the THD is the second harmonics alone. 20 x log(1.4/0.9e-6) -9 = 114.8 dB. The rest of them didn't sum up to anything interesting. But they could have...
interessting application of a locking amp. great video as usual. I own a EG&G Instruments 7260 DSP which should also be able to do such measurement. Now placing the DS360 on the wish list ...
You can use pro audio stuff for high dynamic range at these frequencies. Lately ES chips like the ES9822pro improved the landscape, resulting in devices measuring around 125 to 135dB THD with a noise floor better than 150dB (depending on the device YMMV). And of course it's trivial to connect these to a PC for FFT and see the entire spectrum.
Might be cool to properly measure those specifications. The device may have that performance, but it is possible the board implementation does not. What I see in the ES9822pro's datasheet is -118 dB THD in mono mode. Nothing close to what you are suggesting.
YIKES!!! It's not often we see TSP playing down near MY end of the frequency spectrum (i.e. Almost DC) We didn't even see a 1mm, $20,000.00 connecting cable being used! Are you feeling alright Shariar?
Great Video. Although, -110dB THD equals 0.0003%. I think there's an error on your first picture.. 0.000002% THD would be -154dB, which no generator is able to achieve. I did AC-Susceptibility measurements / Harmonic Analysis on Hi - Tc - Superconductors about 30 years ago. Other scientists stacked several LI amplifiers to collect the higher harmonics simultaneously. Very costly.. Instead, I built a Lo THD FFT system using the ultra-linear HP3458A, which I ordered in 1989, i.e. one of the first ones. With 100kHz / 16bit I achieved -96dB SNR / THD, using a Krohn Hite 4400A analog low distortion generator with specified 0.001% THD. At 50kHz, 18bit resolution, I could demonstrate even -108dB THD, @ 195Hz, if I remember correctly. PTB and I think also NIST still use the HP3458A for making ultra precise and low distortion analysis on AC signals, i.e. AC power measurements. You need to synchronize the sampling frequency with the internal oscillator of the HP3458A, so there is a modification to output the internal 10MHz via an optocoupler. The Black Edition 3458A has / had mysterious non equipped pads on its A/D PCB, which could be assembled with a few components for that purpose. 😎
Hello, great video. But why was it necessary to remove the fundamental when using a lock-in analyzer (LIA)? I understand that removing the fundamental is mandatory when measuring the signal energy with wide bandwidth instruments such as AC millivoltmeters. This was done that way in the late 1960s with the HP distortion meters such as the HP 334A. Otherwise the energy in the fundamental (which we do not want to measure) would totally dominate the measured value. But a LIA is the opposite of a wide bandwidth instrument. As you explained in one of your videos, it has an ultra-narrow bandwidth down to a few ten millihertz. So I would have assumed that it can measure the harmonics correctly even if the fundamental is present. The frequency distance between the 2. harmonic and the fundamental ist 775 Hz, much more than a LIA needs for totally ignoring the fundamental. Is this assumption wrong? Greetings and keep up the good work.
Awesome signal generator, which is almost always used in calibrations (just like the 3458). Which dynamic signal analyzer are you getting, also a Stanford like the SR785? That is an absolute banger, I still don't know if there are modern analyzers which can do a real sweep (with autoranging mid-sweep) which are _that_ fast.
In the audio range one can hack together quite impressive test gear, but my favourite piece of audio test gear is easily my Quantasylum QA403. Good enough for the vast majority of tasks and if a sub-ppm THD source is needed there are always fixed frequency wien-bridge oscillators, notch-filters etc. I was going to post a few interesting links but TH-cam won't let me.
@@jimomertz yes , it can go to -160dB, if you use FFT length 1.2M and averaging . -120dB is with lowest averaging and FFT length 256 , with FFT length 1.2M you get easily go to -160dB
Wouldn't the power amplifier (without additional filtering) at the end of the chain increase the THD again? I don't understand why there aren't filters after the power stage to reduce the distortions introduced by the non-linearities of the amplifier.
Great video. At first after the 2015 measurement I expected you to just go with a notch filter and a DSA, which would be the obvious method, but using the lock-in amp to do the measurements was SO much more interesting! 👍
Isn't just beautiful when you make something work again. I would be worried a bit about those opto-couplers forward current being maybe to high? Why would otherwise the led go bad? Or maybe the type of signal going in being to "sharp" rise time...
I have a Wavetek model 112, it's not as good as this instrument. On serious note I want to adjust it but the pots are buried away and you need a riser card. However the PCB connectors are a weird gold flashed individual soldered pins that work like two scissor's, they interlock perpendicular to each other in a 'scissor action' as you push the card down onto a 'motherboard' below. I can't find them anywhere, am I making any sense ? or have you or your fan's below heard of such an animal please ? or even know where I might find such adapters ? If I could source the pins I would make my own riser and one for anybody who wants one !......cheers
With the RTO, which i'm most familiar with, it's possible to apply a pretty deep digital filter on the first harmonic, and i've been able to measure down to 5 μV signals with it, using the math channel and zero clamping, to get rid of drift and 1/f noise...50-100x averaging gaveme 10,000:1 dynamic range, and, we tested that against the fft, and we were able to get 'direct' measurements that are about as dep as the fft. Nice setup you made...btw, some of those resistors that i have came from a ballistics computer, which had a huge diode /resistor / capacitor / inductor signal processor, which even had temeperature control heater resistors on the board...that group of resistors near the output reminded me of that ....they spared no expense in those military computers inn the 1960's!
I’d love to see the circuit diagram of the power amplifier output stages for them to be such a low distortion. Most power amplifiers seem to manage .01% distortion. I thought it would be fascinating to see how they drop this by such a huge amount annd before before the attenuator.
Not that hard to do better than 0.01% distortion in the audio range, actually some off-the-shelf ICs will pretty much do that. Of course one could also go discrete.
The game of, is the pin stuck in the via? Or is there more solder. Frustrating looking at the board lamination composition and just giving a sigh, "going to be one of those boards" thankfully the lifted trace is part of an isolation circuit and not a high frequency matching circuit. Optocouplers are amazing but surprising when expensive lab equipment are brought down by burned out diods.
Your title 0.000002% seems to be a typo. -113dB is about 0.0002%. Forgot to multiply by 100? thd_percent = 100*10^(thd_db/20). I usually calibrate these by direct reading on NI PXI5922 and get something around that number. But it's actually about limit of 5922.
Very convincing assay technique, but having replaced the malfunctioning Opto-Isolator could the now functional unit not be submitted to a qualified calibration lab, or even to the manufacturer to be able to re-insert it into lab grade inventory with a calibration report to back it up?
4 Signal Path Vids in less than 2 weeks? The gods have heard my cries!
LOL
You are a gem for the hardcore electronics enthusiastic... your new videos always makes me smiling...please make a video on how you achieve all these high skilled stuffs... Thank you so much Sir...
Great video, great explanations as always! I'm pleasantly surprised how much this makes sense given that I took the relevant college classes 40+ years ago and have worked exclusively in the digital domain since then.
Really cool video, and a creative way to measure the THD. I am not as familiar with the lock in amplifiers. I am just wondering if it is still possible that the THD may be worse on other ranges and frequencies of the DS360. Perhaps the signal path used for the 1.4V 775hz does not go through the resistors which reached high temperatures?
Amazing piece of kit, nice troubleshooting and an easy problem to repair.
The biggest flex in this video in my opinion is the little "Bench 8" sticker on the table, I wish I had as much benchspace. ;)
A very good video as per. I really like the SRS equipment UIs and industrial design. The notch method of THD measurement seems so obvious... once it's explained. Thank you for doing so.
I know ! accurately mask the fundamental measure and add up the rest !!
Repair, repair and repair again! Shariar, thanks soooo much for your repair videos, I really like those! Waiting for the DSA to be added to your lab!
3458A is also very capable to measure THD, with floor below -110dB at 1kHz or there about. There are some examples available with GPIB code for this application.
Cool video. I'm not familiar with lock-in amps, so I'll have to watch that other video but using a notch and adding gain seems like an intuitive solution to get the measure.
I haven't worked with a lot of analog equipment, but I have written phase accumulators and worked with SIMD math libraries for audio synthesis before. Cool to see a practical DDS application for lab use.
So glad the repair wasn't just another power supply, but alas just an opto . . . You gotta get that rental place to send you another one that they can't fix. But still enjoyable to see how you verified it.
Really nice video.
The main contributor to the THD is the second harmonics alone. 20 x log(1.4/0.9e-6) -9 = 114.8 dB.
The rest of them didn't sum up to anything interesting. But they could have...
Yes, this is typical in single-ended amplifiers.
interessting application of a locking amp. great video as usual. I own a EG&G Instruments 7260 DSP which should also be able to do such measurement. Now placing the DS360 on the wish list ...
You can use pro audio stuff for high dynamic range at these frequencies. Lately ES chips like the ES9822pro improved the landscape, resulting in devices measuring around 125 to 135dB THD with a noise floor better than 150dB (depending on the device YMMV). And of course it's trivial to connect these to a PC for FFT and see the entire spectrum.
Cosmos ADC ftw ;)
Might be cool to properly measure those specifications. The device may have that performance, but it is possible the board implementation does not. What I see in the ES9822pro's datasheet is -118 dB THD in mono mode. Nothing close to what you are suggesting.
@@Darkknight512 I'm quoting measurements from real life implementations
Yes, man
@@tiftikwhich one? Even the Toppind D90 has only -120dB of THD. But ESS ICs do have around -125dB to -135dB of SNR, maybe you mixed the two?
As always, learned something new, very much enjoyed this episode.
YIKES!!! It's not often we see TSP playing down near MY end of the frequency spectrum (i.e. Almost DC)
We didn't even see a 1mm, $20,000.00 connecting cable being used!
Are you feeling alright Shariar?
Frequency in Hz, BNC connectors, this is certainly an oddball in The Signal Path...
Me to ! I had to sit down !
Great Video. Although, -110dB THD equals 0.0003%. I think there's an error on your first picture.. 0.000002% THD would be -154dB, which no generator is able to achieve.
I did AC-Susceptibility measurements / Harmonic Analysis on Hi - Tc - Superconductors about 30 years ago. Other scientists stacked several LI amplifiers to collect the higher harmonics simultaneously. Very costly.. Instead, I built a Lo THD FFT system using the ultra-linear HP3458A, which I ordered in 1989, i.e. one of the first ones. With 100kHz / 16bit I achieved -96dB SNR / THD, using a Krohn Hite 4400A analog low distortion generator with specified 0.001% THD. At 50kHz, 18bit resolution, I could demonstrate even -108dB THD, @ 195Hz, if I remember correctly. PTB and I think also NIST still use the HP3458A for making ultra precise and low distortion analysis on AC signals, i.e. AC power measurements. You need to synchronize the sampling frequency with the internal oscillator of the HP3458A, so there is a modification to output the internal 10MHz via an optocoupler. The Black Edition 3458A has / had mysterious non equipped pads on its A/D PCB, which could be assembled with a few components for that purpose. 😎
Thanks for catching the type. I have fixed it.
Again lucky man :) (only one optocoupler) Great video! THX
Hello, great video. But why was it necessary to remove the fundamental when using a lock-in analyzer (LIA)? I understand that removing the fundamental is mandatory when measuring the signal energy with wide bandwidth instruments such as AC millivoltmeters. This was done that way in the late 1960s with the HP distortion meters such as the HP 334A. Otherwise the energy in the fundamental (which we do not want to measure) would totally dominate the measured value. But a LIA is the opposite of a wide bandwidth instrument. As you explained in one of your videos, it has an ultra-narrow bandwidth down to a few ten millihertz. So I would have assumed that it can measure the harmonics correctly even if the fundamental is present. The frequency distance between the 2. harmonic and the fundamental ist 775 Hz, much more than a LIA needs for totally ignoring the fundamental. Is this assumption wrong? Greetings and keep up the good work.
I'm astounded that the critical filter was on a breadboard ... and still specs were met!!!
Audio frequencies.
You can get away with absolute murder at KHz and low MHz.
@@digitalradiohacker Thanks.
@@liam3284 I thought about some magnetic materials (nickel?) which if used could add non-linearity.
shahriar to this audio frequency pos: "i find your lack of gigahertz disturbing"
PS. Did the heating of the power amp section sort itself out? I'm not sur if you mentioned it?
Yes, it did.
Awesome signal generator, which is almost always used in calibrations (just like the 3458). Which dynamic signal analyzer are you getting, also a Stanford like the SR785? That is an absolute banger, I still don't know if there are modern analyzers which can do a real sweep (with autoranging mid-sweep) which are _that_ fast.
In the audio range one can hack together quite impressive test gear, but my favourite piece of audio test gear is easily my Quantasylum QA403. Good enough for the vast majority of tasks and if a sub-ppm THD source is needed there are always fixed frequency wien-bridge oscillators, notch-filters etc. I was going to post a few interesting links but TH-cam won't let me.
shame I would be interested in how to generate a 1Khz sine with great THD at home !
Very nice , I bought my self brand new AP555 30k$. It measures down to -120dB THD. It has very nice Software .
I thought the Audio Precisions were good down to 160 dB 🤷
@@jimomertz yes , it can go to -160dB, if you use FFT length 1.2M and averaging . -120dB is with lowest averaging and FFT length 256 , with FFT length 1.2M you get easily go to -160dB
I wonder what settings the Keithley used - THD+n ? or just THD ?
THD up to 5th harmonic.
Nice repair!
Thanks!
do you konw about the "E1DA Cosmos ADC" Projekt it promises THD measurements of -116dB
Maybe review SR560 and SR570 and their NIM/SIM modules? Femto has some good amplifiers. Lecroy has a preamplifier.
Wouldn't the power amplifier (without additional filtering) at the end of the chain increase the THD again? I don't understand why there aren't filters after the power stage to reduce the distortions introduced by the non-linearities of the amplifier.
I guess it is easier to make a clean power amplifier but not a power filter.
Is there a name for the method described here using the lock-in amp to measure the THD?
I am not sure, it is something I came up with during a previous lock-in amplifier video.
@@Thesignalpath Thank you, Shahriar!
Would be useful to show the signal fundamental and harmonics in a Spectrum Analyzer with a very low noise floor?
Remember that noise floor is not the issue here. The linearity is. A spectrum analyzer front-end alone will not have this much THD...
@@Thesignalpath And, as you mentioned in the video, the 50ohm impedance factor. Many spectrum analyzers struggle in the kHz range, too.
Great video. At first after the 2015 measurement I expected you to just go with a notch filter and a DSA, which would be the obvious method, but using the lock-in amp to do the measurements was SO much more interesting! 👍
Isn't just beautiful when you make something work again. I would be worried a bit about those opto-couplers forward current being maybe to high? Why would otherwise the led go bad? Or maybe the type of signal going in being to "sharp" rise time...
I have a Wavetek model 112, it's not as good as this instrument. On serious note I want to adjust it but the pots are buried away and you need a riser card. However the PCB connectors are a weird gold flashed individual soldered pins that work like two scissor's, they interlock perpendicular to each other in a 'scissor action' as you push the card down onto a 'motherboard' below. I can't find them anywhere, am I making any sense ? or have you or your fan's below heard of such an animal please ? or even know where I might find such adapters ? If I could source the pins I would make my own riser and one for anybody who wants one !......cheers
With the RTO, which i'm most familiar with, it's possible to apply a pretty deep digital filter on the first harmonic, and i've been able to measure down to 5 μV signals with it, using the math channel and zero clamping, to get rid of drift and 1/f noise...50-100x averaging gaveme 10,000:1 dynamic range, and, we tested that against the fft, and we were able to get 'direct' measurements that are about as dep as the fft. Nice setup you made...btw, some of those resistors that i have came from a ballistics computer, which had a huge diode /resistor / capacitor / inductor signal processor, which even had temeperature control heater resistors on the board...that group of resistors near the output reminded me of that ....they spared no expense in those military computers inn the 1960's!
Remember that a digital notch filter would not help in that situation. The front-end of the instrument is the THD limitation (and of course the ADC).
XLR output says a lot on its own.
I’d love to see the circuit diagram of the power amplifier output stages for them to be such a low distortion. Most power amplifiers seem to manage .01% distortion. I thought it would be fascinating to see how they drop this by such a huge amount annd before before the attenuator.
Not that hard to do better than 0.01% distortion in the audio range, actually some off-the-shelf ICs will pretty much do that. Of course one could also go discrete.
Feedback can be used to help linearize an amplifier, maybe that's an ingredient in the special sauce.
Thanks. I always learn what i did not know,yes?
Good stuff,thanks.
You can use good audio interface card to measure thd in < 48 kHz < 2 Vpp range signals
(For harmonics like -110 dB below main.)
Love the thumbnail!
The game of, is the pin stuck in the via? Or is there more solder. Frustrating looking at the board lamination composition and just giving a sigh, "going to be one of those boards" thankfully the lifted trace is part of an isolation circuit and not a high frequency matching circuit. Optocouplers are amazing but surprising when expensive lab equipment are brought down by burned out diods.
PS I'd LOVE to have one of those.
Very interesting Video! Thank you!
Your title 0.000002% seems to be a typo. -113dB is about 0.0002%. Forgot to multiply by 100? thd_percent = 100*10^(thd_db/20). I usually calibrate these by direct reading on NI PXI5922 and get something around that number. But it's actually about limit of 5922.
It is! Thank you for catching it. I have fixed it.
-114dBc is not 0,000002% THD. It's "only" 0,0002% ...
Thanks, I have fixed the typo. :)
Very convincing assay technique, but having replaced the malfunctioning Opto-Isolator could the now functional unit not be submitted to a qualified calibration lab, or even to the manufacturer to be able to re-insert it into lab grade inventory with a calibration report to back it up?
My Wavetek model 296 is now cowering in SHAME!!!
half a million times smaller! I can hear that 2nd harmonic because I'm a bat.
Then I get jealous. Mine just pops up. :＜0.05 %, 10 Hz to 20 kHz, 1 Vpp
nothing wrong with that.
@@andymouse It's enough for home use.
@@karlbesser1696 yep it's fine, I have about the same specs.
That thing must've gave itself a lobotomy!
always thumb up
I have to say you had a bit of luck finding the fault that fast.... Come on admit it ha ha. I have to say, I could have been there all day.