I will purchase one, please.
Please let me know when you start selling the QA480. I’d like to get one too.
I would like to purchase one QA480 as well. Cheers.
Hi @CurtisIIX, please send a mail to sales with QA480 in the subject and you’ll get an emailed link when orders can be made.
I read through your notes on the QA480 and the effect of the shared supply. What would be the impact of separating the supplies and grounds of the oscillator and the notch filter? Based on some other products doing that you could make the oscillator function as a balanced or single ended source by alternately grounding or floating one output.
As for tuning the way Fluke handles that on some of the precision standards are extra resistors (they do not need to be .01%) and solder bridges.to alter the primary resistor value.
Needless to say I’m interested and will be following this.
That’s a good question. I think for now, in loopback, there probably wouldn’t be much difference now that the errant 2 kHz that was showing up in the notch is gone as that limited the performance. But, once you stick a really high-powered class D DUT in the loop, maybe it would help. Specifically, there are some really high performance class D amps with that are hitting -116 dB THDN and with 131 dB of dynamic range that could conceivably demonstrate the need for additional decoupling.
I wrote a short study on the differences between the common high power class D (such as TPA3255) and this next generation class D (Purifi 1ET400A) HERE. The study looks at noise floor. If there was a situation that might demonstrate the need for more isolation, it’d be that. That amp runs at a 500 kHz switching frequency, so the ~90 dB of attenuation the QA451 has at 500 kHz would help. But I don’t know if it would be enough. The plot below is the SPICE sim of the QA452 filter (from HERE).
QA480 boards went to assembly earlier this week, and software has been added (not released yet) to the QA401 to allow you to automate sweeps of the QA480 into your DUT. A first plot, 3 runs at 256K FFT, is below. You can see where the +12 dB gain stage kicks in at anything > 6 dBV output. On the units being built, output will be within +/- 0.5 dB of indicated. But there will be instructions on how to tweak if you want it closer.
Next week I’ll post THD+N plots and the QA401 software with the new QA480 plugin should be released. And middle of next week emails will be sent with a link to order if you have sent a mail to sales with “QA480” in the title. And hopefully, the following week they’ll go out.
Note the interface to control the QA480 amplitude is REST and very simple. Just do a
HTTP PUT http://localhost:9480/Settings/Level/4
and the QA480 will set its output to 4 dBV (+/- 0.5 dB). Should be a snap to automate from any language. (And just to clarify, the REST processing is done in the app, not on the QA480).
Block diagram (and UI) of the QA480:
I would like to purchase a unit.
The units started shipping yesterday (Friday, 8/22) and I think the remainder of orders will go out Monday and Tuesday of next week. The application for controlling the QA480 is located here.
This is 3 runs on unit MEE for THD+N. Best case -126 dB at 15 dBV output level.
This is 3 runs on unit MEE for THD. Best case about -146 at 3 dBV output.
My unit arrived just now, all the way on the opposite coast. (Somehow the USPS person missed us yesterday.)
I know this is a special offering, but just how do you sell these at the price and still make any money? I haven’t opened the unit (or even plugged it in) but I can see that the bottom of the chassis may be a stock design you already use, but the top has a custom silkscreen and the front panel is unique to this product. Obviously, the PCB is as well. All rather amazing considering the volume.
Everything is built with the aim of maximizing re-use and minimizing part count. The former saves engineering effort, the latter improves reliability, reduces validation needs and reduces assembly cost.
The bottom and top cases are custom and common to all products (they are sourced from China). The QA480 front panel is re-used from the QA351. But even still, the front panels are all laser cut in Utah or Nevada and then powder coated in Washington State.
The USB section of the QA480 is re-used from the QA451, which borrowed from the QA351. The split rail supply is common throughout many of the products. The relay drive section was borrowed from the QA490.
And the case markings are done with a UV printer. This is like a inkjet printer, except it squirts drops of ink, and then exposes the ink to a super bright flash of UV light, which turns the ink into a hard plastic. It’s not as sharp as silk screen, but 50 cases can be printed in an hour, and the artwork can be changed as often as needed. The printers are expensive, starting around $12K to $35K (China versus domestic). But they are worth it.
I don’t think the aesthetic design of the products will ever win an award. But the flexibility cannot be beat.
PS. Please feel free to sweep your THD and THD+N as outlined on WIKI PAGE and share it here. I think it would be very helpful to see the full plots of the different units. It takes a long time to run (about 8 minutes per plot).
Thanks - I’ll try the THD and THD+N sweeps, hopefully sometime in the next week or so. Despite being able to work from home and therefore having no commute at all, it’s amazing how many other things crowd the schedule. :8^)
I also want to try using the Crystek oscillator I bought a while back. Doing a back to back test with the same clock for the A/D and D/A is sort of cheating, but comparing the stock clock to the Crystek using an external oscillator is fairer game. That’s the justification I’ve used for not getting that task done, anyway.
Your approach to building these units is really encouraging. Even home builders of one-of’s could potentially use the methods you employ to make better looking devices than just drilling a panel with a hand drill and applying stick-on labels. Yeah, the gear may not look as swanky as Boulder amplifiers (just to pick one I was looking at last night), but one can’t have everything unless you want to spend big amounts or you own your own CNC machine. The QA products look just fine.
Getting PCB’s made in small quantities is not as daunting or as expensive as it once was. (Note: If anybody has a recommendation for a reliable and reasonably priced source for Rogers RO4350 boards, please let me know.) The tougher one is still the metal work. That may be my own lack of experience in that area showing, but nice housings for test or audio gear doesn’t seem to grow on trees.
Since you started the topic of the front panels manufacturing, another rather affordable option is printing on anodized aluminum. It’s not very DIY friendly, but it’s possible to find a local manufacturer that will do small batches of front panels. They can sometimes do press-in nuts and stand-offs.
My local fab can do as little as a single sheet (IIRC 350x500mm, depends on the thickness), and you can mix and match different designs. The print is very fine and endurable. One-offs are somewhat more expensive, but if you have several people, this is a very good option.
I received the unit on Friday - and have done the first test measurements as instructed on the wiki page. I just wanted to share the results and the it clearly indicates the unit deliveries as promised.
Here’s my results.
I obviously need to get a better 5VDC supply for the USB hub. Or something. I tried not using the hub. Pretty. much the same, just a little different. I tried unplugging everything in the vicinity (except the computer). No difference. All magic solutions considered.
Hi @BKDad, I think the 60 Hz you are seeing likely radiated into the setup. In addition to unplugging things as you’ve done, does the QA401 box show any situational change? That is, if you rotate it 90 degrees on the desk and re-measure are the 60 Hz spurs changing in amplitude? Your 60 Hz is showing up around -132 dBV, which is 250 nVrms, so it could be something that isn’t actually nearby (iow, a high current motor located further than you might expect).
Also, does the spur change if you increase the osc attenuation from 0 dB to 10 dB? If the spur drops by 10 dB, then that means the 60 Hz is coupling into the oscillator. If you short the input to the notch does the spur go away? And finally, if you short the input to the QA401 are you getting a clean spectrum? Try to find which stage this is creeping in.
I tried moving the QA4xx boxes, rotating them, and turning them sideways. Only minor effects.
All the motors in the house are located at least 20 feet away, at least as far as the magnetic waves fly.
I have a list of different things to try when I get a chance. Not panicking yet, at all. I’m certain it can be resolved with a simple solution. It’s a question of finding it and implementing it.
Thanks for the ideas - I’ll try all of them, too.