The link below covers some details on the QA402.
Are you going to be offering any upgrade or trade in allowance for those of us with the QA401?
Given USB hubs like the āAnker 7-Port USB 3.0 Hubā, 36 USD, which has a 1.5A port, it would be a shame to see available power limit the design in any way. I have many of these, or similar plugged in all the time.
Is the screen shot of the QA480 along with the QA402 made through the notch filter or directly? Iām guessing the latter, based on the distortion level.
My bug question is: Why is the base of the 1 KHz QA480 so wide? That just the measurement? Or, something to do with the 1/f noise of the new opamps? The conversion clock? (I donāt see a place on the pcb for an alternate user-supplied Crystal clock as there is on the QA401ā¦)
Overall, nice work!
Is the screen shot of the QA480 along with the QA402 made through the notch filter or directly? Iām guessing the latter, based on the distortion level.
Yes, this is a direct measurement. The QA480 is used as nearly zero distortion reference (no notch). But the notch is useful for understand the DAC and isolating any DAC/ADC interaction
My bug question is: Why is the base of the 1 KHz QA480 so wide? That just the measurement? Or, something to do with the 1/f noise of the new opamps? The conversion clock? (I donāt see a place on the pcb for an alternate user-supplied Crystal clock as there is on the QA401ā¦)
Itās a function of window and FFT size. In the plot following on the post you see a narrower skirt. The clock on the QA402 has typical jitter numbers of 3ps RMS period jitter and phase noise at -125 dBc/Hz at 1 kHz offset. The QA401 clock was 5 ps period jitter. The Crystek at 1 kHz is -148. As the design is fully synchronous and lock to the reference crystal (as was the QA401) I think the jitter is largely inconsequential.
Are you going to be offering any upgrade or trade in allowance for those of us with the QA401?
Nothing has yet been planned. First goal is to get a solid replacement in place for the QA401. The QA402 wonāt have the dynamic range of the QA401, but given the relays, it should outperform the QA401 at most all measurement points.
Given USB hubs like the āAnker 7-Port USB 3.0 Hubā, 36 USD, which has a 1.5A port, it would be a shame to see available power limit the design in any way. I have many of these, or similar plugged in all the time.
Yes, agree. This hub below is 2.5A total for US$20. They say not to use more than 300 mA per port, but upon opening it and looking at the PCB itās a straight shot from the DC inlet to the USB connectors. No PTC or anything. And the power traces on the PCB look adequate too. The QA402 reports 4.88V when plugged into this hub.
āAs the design is fully synchronous and lock[ed] to the reference crystal (as was the QA401) I think the jitter is largely inconsequential.ā
Iām not sure thatās true when the source - like the QA480 or a DAC - has no clock connection to the QA401 or QA402.
Iām being a baby, eh? 
Hi BK, If both boxes are locked to a crystal internally, then the jitter between a DAC and QA401 is set by the worst-case crystal performance on either device (assuming one is much worse than the other). Someone recently shared some discussions in email on folks that are building phase noise measurement hardware for sound cards. Really interesting stuff.
Exactly! Which is why the option for a better internal clock is always good.
Phase noise measurement at between the equivalent of Audio Precision and QuantAsylum levels (in general terms of sophistication and cost) is not for the faint of heart.
Hello matt, wery nice āupgradeā , especially input attenuator is great added value ! What is the expected date of product being available and price range ? Thx, Petr
Hopefully March 2021 and at a similar price to the QA401. As the post indicates, features may be added/subtracted before that time.
It looks like the Crystek CCHD575 is footprint compatible with the current crystal used in the QA402. This gives
| Osc | Footprint | PN @ 1kHz | PN @ 10kHz | PN @ 100 kHz | Comment |
|---|---|---|---|---|---|
| CCHD957 | 14x2x9.2 | -149 | -162 | -170 | QA401 Option |
| CCHD575 | 5x7.5 | -143 | -154 | -162 | QA402 User Upgrade |
| CWX813 | 5x7 | -125 | -140 | -145 | QA402 Stock |
The QA401 had a footprint for the CCHD957 if someone wanted to place it themselves (Iām not aware of anyone doing that and reporting a change in measurements), but the QA402 wonāt have the '957 footprint. The CCHD575 is pretty close to the CCHD957 (within 6-8 dB) and is footprint compatible with the oscillator used in the QA402 (CWX813). The CCHD575 is 15-20 dB better than the stock oscillator the QA402 will use. So, if someone absolutely positively had to have a better clock, they could swap in a CCHD575. But Iād be very, very surprised if it changed what was measured.
Remember, most jitter measurements you might read about (done with J-Test) are related to AES/EBU/SPDIF or a design that is severely broken in some way. The jitter in those articles is related to the transport and is many orders of magnitude worse than the jitter related to the clock jitter in a purely synchronous design. To see the jitter of what is being discussed above would require a dedicated test box that measures directly at the crystal frequency. Iām pretty sure it cannot be measured at 1 kHz (but could be wrong). In other words, I donāt think jtest stands a chance of measuring a difference between the clocks noted above. Can anyone else confirm/refute?
1 Like
Hi Matt,
Thank you for the update. +36dBV is tantalisingly close (but āno cigarāā¦) to +40dBV which is awesome for testing 100V-Line amplifiersā¦
Sorry to hear about the AKM problems.
1 Like
Hi @mark2c, it might be possible. I think the issue might be more with ensuring a CAT I 100Vrms rating would be able to cope with ~600V surge. But it seems itād be pretty easy to make an external high-z divider too. Do you offhand know typical noise levels and THD levels on the 100V side of things?
1 Like
Thanks @matt,
Distortion is typ 0.2% (transformer contribution dominated) and the noise -110dB relative to 100Vrms.
This is a very good protection circuit FYI (low capacitance and no leakage so no contribution when off, sharpest ākneeā when starting to conduct and cheap).
Also, the new gas discharge tubes have come of age with ultra-low capacitance and a 20% accuracy on firing voltage: GTCS35-750M-R05-FT-2.
I have been down the rabbit hole of different clocks on an ADC. It can show up and sometimes embarrassing ways. The best demo was the AK5394A demo board. It used a logic gate and crystal as its master clock. Not bad but swapping for a clean external clock (CCHD957 I think) got another 10 dB in THD and an improvement on SNR. It was easy to see the clock modulation in the spectrum of a clean analog clock sourceā¦
Now I have something to explore with the QA401. . .(Where did I put the extra clock chips. . .)
If you go linux too Iām all in 
Hi Matt. I find the new qa402 interesting, especially the very versatile input attenuator section. What leaves me doubtful is the choice of XLR connectors for input and output. I find that the solution adopted in the qa401 is very versatile: the use of BNC connectors greatly simplifies the connections without the use of adapters, the X 1 probes of the oscilloscope can be used directly without any complication.
1 Like
Hi @Claudio, I also love BNC connectors. The important piece of the puzzle is the connector shown below. This goes from BNC to XLR, and it handles the shorting of the unused input IF you want to single ended (this is from Amazon). This would be used on the inputs.
If you want fully diff input broken out, then XLR to RCA is common (this is from Amazon)
And if you wanted to use two scope probes on L+ and L-, for example, youād use the XLR to RCA, use a gender changer on the RCA, and then use RCA to BNC. On the QA401, the spacing of the BNC was too tight for most of scope probes.
The big sticking point was XLR on the output. Since the XLR type of single ended adaptor shown above shorts one of the output legs, it is bad to connect it to the outputs (as one leg of the output will be shorted and the shorted opamp can get very, very hot and damage itself). And thatās why the BNC were added to the outputsāso that you could have easy access to a single ended signal.
So, while BNC are great connectors, itās a lot of tedium to plug/unplug stereo diff pairs and they donāt leave a lot of real-estate on the front-panel for descriptive graphics. The XLR I/O should eliminate a lot of that tedium, save real-estate, and the input adapter to get back single-ended BNC is cheap and plentiful.
1 Like
Hi Matt. I have another question about the new qa402. Since the current consumption of the qa402 will greatly exceed the current capacity of the usb sockets making it necessary to have a powered usb hub, why not provide a Power Jack connector on the QA402 in such a way as to connect a simple and economical power supply (they are easily found on ebay or on amazon) external on the qa402 so as not to have to use any usb hub? Thanks for the reply