That is one analyser I always wanted to play with. Never had a chance to use one. An 8903B would be pretty cool.
I've used the Motu M4 and M6 for various measurements in 192k and have seen good results. I tried the 2i2 and similar Focusrite boxes and wasn't thrilled with the noise floor or upper frequency cutoff limit. They're also not well behaved with the PC based platforms.
I've rented AP analyzers in the past for commercial work, as most of the upscale studio customers I worked with insist on repeatable results. I hate the subscription based software thing most companies are doing now. When Avid started this, I stopped dealing with them. Their prices were already insane and alot of hardware wasn't PC compatible.
For speaker stuff, REW, ARTA and Clio have done well for my needs. The Motu boxes perform very well using the built in preamps and Earthworks mic. Even on USB power, they do well. The M6 requires external power and therfore does a little better above 40 kHz under -90 dB. The M4 is perfect for portable measurements as it doesn't need to go past 25 kHz.
For doing basic bench work ie. HD measurements on amps, tape machines and other small signal audio applications, I've used software developed by a guy I knew who worked for HP back in the day. With the M6 preamps and converters, I've got a decent noise floor to work with. I don’t need .001% resolution. I have an RME ADI Pro2 FSR which has lots of filtering options, but even thats overkill for what I'm doing.
I've rented AP analyzers in the past for commercial work, as most of the upscale studio customers I worked with insist on repeatable results. I hate the subscription based software thing most companies are doing now. When Avid started this, I stopped dealing with them. Their prices were already insane and alot of hardware wasn't PC compatible.
For speaker stuff, REW, ARTA and Clio have done well for my needs. The Motu boxes perform very well using the built in preamps and Earthworks mic. Even on USB power, they do well. The M6 requires external power and therfore does a little better above 40 kHz under -90 dB. The M4 is perfect for portable measurements as it doesn't need to go past 25 kHz.
For doing basic bench work ie. HD measurements on amps, tape machines and other small signal audio applications, I've used software developed by a guy I knew who worked for HP back in the day. With the M6 preamps and converters, I've got a decent noise floor to work with. I don’t need .001% resolution. I have an RME ADI Pro2 FSR which has lots of filtering options, but even thats overkill for what I'm doing.
Some 15-or-so years ago, a lot of used HP 8903 showed up on ePay in various states of disrepair. I bought 5-6 of them over time, repaired them, ran them through the performance test in the service manual, and sold them. Made a little beer money, but mostly had fun. They were neat analyzers. They were very modular in construction.
They were also hard to destroy. I accidentally connected the input to the screen grid of an output tube. All that happened was that the 64 mA (!) fuse in the input blew.
Tom
Hi Tom,
I'd like to run into one that needed TLC today. People want the moon for everything. I like to restore test equipment and have a small collection, along with original manuals and catalogues. Yeah, I'm one of those! lol!
I'd like to run into one that needed TLC today. People want the moon for everything. I like to restore test equipment and have a small collection, along with original manuals and catalogues. Yeah, I'm one of those! lol!
Hello All,
In the cage I have a APx555 with the current APX500 software, a APx1701 transducer test interface (amplifier, microphone input tool .....) and an assortment of GRAS microphones.
What I want to do is spinorama type speaker test plots. I have sketched up a plan for a manual speaker positioning turntable. I suspect that I can save the the raw Audio Precision data and export it to some other software to process the spinorama plots.
Recommendations for the spinorama software? Does REW or other downloadable software do this?
Thanks DT
In the cage I have a APx555 with the current APX500 software, a APx1701 transducer test interface (amplifier, microphone input tool .....) and an assortment of GRAS microphones.
What I want to do is spinorama type speaker test plots. I have sketched up a plan for a manual speaker positioning turntable. I suspect that I can save the the raw Audio Precision data and export it to some other software to process the spinorama plots.
Recommendations for the spinorama software? Does REW or other downloadable software do this?
Thanks DT
Last edited:
I like test equipment too. Thankfully my storage space is limited.
I seem to recall paying about $150 shipped for the broken 8903s. But that was 15+ years ago. Most just needed new power supply capacitors (screw-in types).
Tom
Spin-o-rama is ill-defined as it just says something is rotating ;-)
Depending on what plot formats you want to have various software choices come to mind.
I've use VACS for years if not decades (the viever is free of charge), but REW and VituixCad are viable free options these days.
If you want a high-end HP analyser to repair, this could be your lucky day
Do you feel lucky?
NB - the seller is the fellow Helmut refers to. You will see a plethora of other AAs in his listings. He doesn't sell them cheaply. His website lists a couple of Wandel and Goltermann AAs. Modern test gear doesn't give me G.A.S, but these do. My concern is what happens with 50Kg of CRT-loaded analyser in the courier system.... (and the fact that I've seen them for under half the price, although I doubt the condition was as good as our man in Dusseldorf. I thought I had a fetish for AAs, but this guy takes the biscuit...)
Hi thermionic,
Many thanks!
I just replaced all my doors and windows in the house, so I am currently broke. I find myself considering it, but the value of the Canadian dollar kills the deal. Even shipping in Canadian funny money is very high.
I wish!
Many thanks!
I just replaced all my doors and windows in the house, so I am currently broke. I find myself considering it, but the value of the Canadian dollar kills the deal. Even shipping in Canadian funny money is very high.
I wish!
That all depends on how well the seller packs the analyzer. If they use a foam-in-place machine the gear will be fine. I had an HP 3577A shipped to me using that method. Sure. The cardboard box took a beating, but the analyzer was fine. The 3577 weighs about that much.
Tom
Indeed. I've collected vintage scopes without issue. However, in some cases they were really badly packed and it was a miracle that they didn't get smashed. I have seen that otherwise intelligent people can often pack stuff really badly. I got a tape machine from a doctor and it was rattling around in the box. The whole chassis was warped...
I suspect the guy in Germany will pack the W+G well. I'm not about to get it, as his price is over double what I've seen them for and it would be pure G.A.S anway. I would still use the UPV as a workhorse and the W+G would be for nerding out... (which I spend waay too much time doing anyway...).
I suspect the guy in Germany will pack the W+G well. I'm not about to get it, as his price is over double what I've seen them for and it would be pure G.A.S anway. I would still use the UPV as a workhorse and the W+G would be for nerding out... (which I spend waay too much time doing anyway...).
Hi thermionic,
I bought my R&S UPL from Knopf in 2017. The device has been running here for almost 8 years without any problems. The analyzer is really well-built and robust. (Runs with Windows 98 and boots in < 40 seconds).
The only drawback is that I can't perform hardcore THD measurements below -140 dB with it. I already wrote that it becomes inaccurate at < -125 dB THD. However, unlike AP, it can at least display the THD, namely all harmonics K2-K9. You also know that from the UPV. I was at Karl Knopf in Düsseldorf 5 months ago, and we measured one of my oscillators with his APx555B. I didn't believe him at the time that this €60,000 device didn't display THD or individual harmonics in a live FFT. I think this is a bad joke. Should I now calculate the THD from the FFT using a calculator?
Best regards,
Helmut Sell
I bought my R&S UPL from Knopf in 2017. The device has been running here for almost 8 years without any problems. The analyzer is really well-built and robust. (Runs with Windows 98 and boots in < 40 seconds).
The only drawback is that I can't perform hardcore THD measurements below -140 dB with it. I already wrote that it becomes inaccurate at < -125 dB THD. However, unlike AP, it can at least display the THD, namely all harmonics K2-K9. You also know that from the UPV. I was at Karl Knopf in Düsseldorf 5 months ago, and we measured one of my oscillators with his APx555B. I didn't believe him at the time that this €60,000 device didn't display THD or individual harmonics in a live FFT. I think this is a bad joke. Should I now calculate the THD from the FFT using a calculator?
Best regards,
Helmut Sell
Crikey! I always thought it odd that R+S gave you live FFT in the early 90s with the UPL, yet AP only offered a snapshot FFT right up until the 500-series in the 2010s... The feature that I would miss the most from the UPV is the random generator / live FR FFT ability. As a pro, I want to have the best tools and the 555 has the lowest THD+N. My concern is that it will see less use than the UPV purely because of this feature. If you need to align knobs for an EQ it's seriously useful. Likewise, you can check channel balance on pots far easier than when using single or multi tones - you simply watch flat lines as you manipulate the pot. It's such a cool feature.
I think I might start a separate thread about whether the input section on the UPV can be improved. It uses relatively exotic parts such as the OPA627 and AD797 which weren't around when the UPL was released, yet its THD+N is only a fraction better (Karl's listings quote the UPV at -114dB and UPL at -113dB, but units will vary - we have 4 UPVs and there is around 3dB variation, i.e. best can nudge -115, but worst struggles to get -112). If R+S deployed parts in the UPV that weren't around when they designed the UPL, then I worry that the ADC is the limiting factor... It has a bunch of NE5534s for the ranging mechanism, but I am not sure that there is an obvious IC that betters them in this duty for noise (THD, yes - there must be). It also has OPA2134 and OP275, but I would bet they are servos (I cannot imagine the OP275 would be in the path of any AA as it's inferior to a 5532 in many areas and only betters it in terms of speed). Most of the signal tracking is inner-layer, so it will need very, very careful tracing (I think I'd have to put a signal through it and trace it with a probe - one slip with the probe and it could be major damage...).
I think I might start a separate thread about whether the input section on the UPV can be improved. It uses relatively exotic parts such as the OPA627 and AD797 which weren't around when the UPL was released, yet its THD+N is only a fraction better (Karl's listings quote the UPV at -114dB and UPL at -113dB, but units will vary - we have 4 UPVs and there is around 3dB variation, i.e. best can nudge -115, but worst struggles to get -112). If R+S deployed parts in the UPV that weren't around when they designed the UPL, then I worry that the ADC is the limiting factor... It has a bunch of NE5534s for the ranging mechanism, but I am not sure that there is an obvious IC that betters them in this duty for noise (THD, yes - there must be). It also has OPA2134 and OP275, but I would bet they are servos (I cannot imagine the OP275 would be in the path of any AA as it's inferior to a 5532 in many areas and only betters it in terms of speed). Most of the signal tracking is inner-layer, so it will need very, very careful tracing (I think I'd have to put a signal through it and trace it with a probe - one slip with the probe and it could be major damage...).
Hi thermionic,
With analog oscillators that have R and C as frequency-determining components, you have to dimension the resistors as low as possible to get the THD+N into the -120 dB range. For example, in the range of 1-3 kOhm. However, you can't choose these much lower, because the THD of the op amps used in the oscillator circuit would then increase too much. Therefore, trying to replace the op amps probably won't help. For an R-C oscillator with a THD+N of, say, -113 dB at 1 kHz/22 kHz Bw, you would have to reduce the R values by a factor of 2 to achieve a THD+N improvement of about 3 dB. But then you would only be at -116 dB, and you would still be missing 4 dB. If you swap the op amps, for example, using an LM4562, OPA1612, or OPA1656, you might be able to achieve another 1 dB improvement. But probably not 6-8 dB. There are generally three problems here, which are independent of the oscillator configuration used (Wien bridge, bridged-T, twin-T, SVO, etc.):
1) R values in the frequency-determining network as low as possible.
2) Optimal amplitude control that does not inject additional noise into the oscillator at the resonant frequency.
3) All other components that follow the oscillator core, such as buffers, voltage dividers, and subsequent amplifiers, must be dimensioned for ultra-low noise.
If you now offer a huge frequency range for your analog oscillators, such as the R&S Option B1 >100 kHz or the AP >200 kHz, additional design criteria are added to ensure stability.
So, if you don't consistently implement an ultra-low noise+distortion approach in all stages of an oscillator during the design phase, I don't think you'll achieve much by swapping out a few op amps. You'd have to redesign the whole thing.
Best regards,
Helmut Sell
With analog oscillators that have R and C as frequency-determining components, you have to dimension the resistors as low as possible to get the THD+N into the -120 dB range. For example, in the range of 1-3 kOhm. However, you can't choose these much lower, because the THD of the op amps used in the oscillator circuit would then increase too much. Therefore, trying to replace the op amps probably won't help. For an R-C oscillator with a THD+N of, say, -113 dB at 1 kHz/22 kHz Bw, you would have to reduce the R values by a factor of 2 to achieve a THD+N improvement of about 3 dB. But then you would only be at -116 dB, and you would still be missing 4 dB. If you swap the op amps, for example, using an LM4562, OPA1612, or OPA1656, you might be able to achieve another 1 dB improvement. But probably not 6-8 dB. There are generally three problems here, which are independent of the oscillator configuration used (Wien bridge, bridged-T, twin-T, SVO, etc.):
1) R values in the frequency-determining network as low as possible.
2) Optimal amplitude control that does not inject additional noise into the oscillator at the resonant frequency.
3) All other components that follow the oscillator core, such as buffers, voltage dividers, and subsequent amplifiers, must be dimensioned for ultra-low noise.
If you now offer a huge frequency range for your analog oscillators, such as the R&S Option B1 >100 kHz or the AP >200 kHz, additional design criteria are added to ensure stability.
So, if you don't consistently implement an ultra-low noise+distortion approach in all stages of an oscillator during the design phase, I don't think you'll achieve much by swapping out a few op amps. You'd have to redesign the whole thing.
Best regards,
Helmut Sell
Thank you, Helmut. You make some excellent points. From my point of view, there are few high performance oscillators out there now. Of course, there is a lot of variation in terms of flexibility. Some are 'one trick ponies' that achieve low THD+N at the expense of convenience. The AP and R+S oscillators are the work of large and well-resourced design teams, which is reflected in their performance / flexibility ratio.
One of my UPVs is a 'beater' (it's the one I showed photos of earlier - I would be reluctant to do any work on the others as they are used in my business). My initial focus would be the input stage. I am hoping to get a couple of Victor's oscillators soon, so that I can see just how much of the distortion is inherent to the I/P stage... Compared to the B1, with the 2722 oscillator I get slightly more 3rd harmonic, but better overall THD+N by ~ 1dB. I know a couple of people that have 555s, but they are a long way from me, so a lot of work to drag a UPV over to them (and they both have the non-B versions, which according to Sam G's paper, can have higher THD oscillators than the 2722).
Despite the 5534 qualifying as a 'vintage' part, my suspicion is that the limiting factor in the UPV's input stage will be the ADC... Uprating the PSU and swapping the 5534s for more modern parts may yield improvement, but I suspect it is unlikely to be more than a dB or two. I hope I am wrong!
One of my UPVs is a 'beater' (it's the one I showed photos of earlier - I would be reluctant to do any work on the others as they are used in my business). My initial focus would be the input stage. I am hoping to get a couple of Victor's oscillators soon, so that I can see just how much of the distortion is inherent to the I/P stage... Compared to the B1, with the 2722 oscillator I get slightly more 3rd harmonic, but better overall THD+N by ~ 1dB. I know a couple of people that have 555s, but they are a long way from me, so a lot of work to drag a UPV over to them (and they both have the non-B versions, which according to Sam G's paper, can have higher THD oscillators than the 2722).
Despite the 5534 qualifying as a 'vintage' part, my suspicion is that the limiting factor in the UPV's input stage will be the ADC... Uprating the PSU and swapping the 5534s for more modern parts may yield improvement, but I suspect it is unlikely to be more than a dB or two. I hope I am wrong!
A couple fixed frequency oscillators are actually not a negative. I have a couple, 1KHz and 10 KHz Victor's oscillators. They work as advertised.
Yes, everything from the power supply, case and everything else becomes important for oscillator performance at the levels you are talking about.
Yes, everything from the power supply, case and everything else becomes important for oscillator performance at the levels you are talking about.
I'm for a 'cheap' system based on available software that is well supported, and continues to advance in technical and functional performance, especially when coupled with a ubiquitous hardware device like a soundcard, whether that be a vintage EMU or stock Focusrite or future 384kHz sample rate widget, and a battery powered laptop.
There may be a hope that REW could be automated to capture the harmonic presets that minimise THD for a particular frequency and level of test signal for a loopback configuration, and then apply those presets for an external test configuration, as part of a test campaign (such as a step frequency plot). Such a setup could suppress harmonic levels into the noise floor of the soundcard. My concern would be that REW likely needs to either action, or hold and return from, a different external configuration (ie. a loopback alternating with a test configuration), and that may not be possible. But if it was, and John deemed it worthy of REW, then I'd anticipate a substantial improvement in default cheap performance.
There may be a hope that REW could be automated to capture the harmonic presets that minimise THD for a particular frequency and level of test signal for a loopback configuration, and then apply those presets for an external test configuration, as part of a test campaign (such as a step frequency plot). Such a setup could suppress harmonic levels into the noise floor of the soundcard. My concern would be that REW likely needs to either action, or hold and return from, a different external configuration (ie. a loopback alternating with a test configuration), and that may not be possible. But if it was, and John deemed it worthy of REW, then I'd anticipate a substantial improvement in default cheap performance.
This is my solution for Measuring Distortion on the Cheap:
I modified the Sound Blaster X-Fi HD (SB1240) sound card. Replaced the Cirrus Logic CS5361-KZZ to CS5381-KZZ ADC chip, it is pin compatible. Next I replaced the TI MC33078 dial operational amplifiers in the recording path to OPA1656. I replaced the reference voltage filter capacitor of the ADC to bigger (can't remember the value). Then I bypassed the ISL54405 analog switches from the RCA inputs to the opamp inputs by 33uF/50V capacitors. This way I lost the MIC and Phono input functionality, but I use just the Line input anyway. The optimum input level (lowest THD) was found 200 mV, overload limit 440 mV. Probably I have to build a passive attenuator for measuring higher level signals. See loopback measurement.
I modified the Sound Blaster X-Fi HD (SB1240) sound card. Replaced the Cirrus Logic CS5361-KZZ to CS5381-KZZ ADC chip, it is pin compatible. Next I replaced the TI MC33078 dial operational amplifiers in the recording path to OPA1656. I replaced the reference voltage filter capacitor of the ADC to bigger (can't remember the value). Then I bypassed the ISL54405 analog switches from the RCA inputs to the opamp inputs by 33uF/50V capacitors. This way I lost the MIC and Phono input functionality, but I use just the Line input anyway. The optimum input level (lowest THD) was found 200 mV, overload limit 440 mV. Probably I have to build a passive attenuator for measuring higher level signals. See loopback measurement.
Attachments
Due credits and a big thanks should go to @googlyone for inspiration:
https://www.diyaudio.com/archive/bl...e-sound-balster-x-fi-music-sb1240-puzzle.html
https://www.diyaudio.com/archive/bl...e-sound-balster-x-fi-music-sb1240-puzzle.html