Sorry if this has been posted already (google did not turn up anything):
Sneak Peek: The QA402
– QuantAsylum
They had to change to TI DAC and ADCs due to the fire at AKM.
My impression so far was that the AK5397 is the champion for noise but not distortion (that was the discontinued AK5394A) and that the PCM4220 didn't even come close. Similarly, the PCM1794A is good, but not the very best DAC around.
Yet they changed to these components and claim to have achieved even better noise performance than before. They also show some distortion measurements with a -25 dB carrier and loopback, so it is hard to tell how good they are compared to the older QA-401.
Sneak Peek: The QA402
– QuantAsylum
They had to change to TI DAC and ADCs due to the fire at AKM.
My impression so far was that the AK5397 is the champion for noise but not distortion (that was the discontinued AK5394A) and that the PCM4220 didn't even come close. Similarly, the PCM1794A is good, but not the very best DAC around.
Yet they changed to these components and claim to have achieved even better noise performance than before. They also show some distortion measurements with a -25 dB carrier and loopback, so it is hard to tell how good they are compared to the older QA-401.
The lower noise comes from the DAC side of things. The 1794 is quieter than the 4490.
The 4220, while not a 5394 for distortion, or a 5397 for noise, is more in the middle. Turn the input level down to around -20dB and the distortion drops below -120dB. Same as most modern ADCs. The 1794 also reaches a distortion minimum around this output level too.
I always thought that the old 401 was a decent product at an attractive price point, but it had it's limitations. The 402 is basically the same with a different bunch of limitations.
The 4220, while not a 5394 for distortion, or a 5397 for noise, is more in the middle. Turn the input level down to around -20dB and the distortion drops below -120dB. Same as most modern ADCs. The 1794 also reaches a distortion minimum around this output level too.
I always thought that the old 401 was a decent product at an attractive price point, but it had it's limitations. The 402 is basically the same with a different bunch of limitations.
Thanks. I once spent some time tracking down this info that is not really contained in the data sheets here in the forum, but I didn't pay much attention to either TI chip because I wrongly assumed that they were not in the same league.
So are you saying the 4220 is between the the AKMs both for noise and distortion?
How does the 1794 compare to the 4490 or newer parts like 4493 and 4497 for distortion?
So are you saying the 4220 is between the the AKMs both for noise and distortion?
How does the 1794 compare to the 4490 or newer parts like 4493 and 4497 for distortion?
The 1794 doesn't really reach the same distortion minimum as these newer ones can. They will dip down into the 0.0000x% territory. I've seen the 1792/4 going into 0.0001x% territory though, which is still very low by most standards. It's low enough for the vast majority of things, readily available and not madly expensive.
For cost effective measurement gear the 4490 hit a sweet spot though of being easy to implement and inexpensive.
The interesting thing about the 1792/4 is their digital filter in sharp mode. This is, aside from its latency, still one of the best in any of the new DAC chips. It has 130dB of stop band attenuation, which completely mitigates any aliasing products within the devices signal to noise ratio. It's also got very low passband ripple. Most modern ones come with 120dB or, in the AK4499s case, only 100dB. The AK4499 has been highly criticised on this as aliasing products creep in and affect its performance. Certainly not in any major way, but it's still a point that could be better. If TI could do it almost 20 years ago why can't you?
AKM know about this too and in their AK4498 they've included a filter with 150dB stop band attentuation to completely eliminate aliasing products within the devices signal to noise ratio.
This is a minor point and in most testing scenarios it wouldn't be an issue, it's just not as good as it could be. The 1792/4 won't have these problems though.
Individuals have posted measurements of the top TI ADCs delivering 0.0000x% distortion when the input signal is reduced from maximum, better than spec would indicate. In a way this is similar to the AK5394. If you go by the data sheet AKM provide it's performance is good but not outstanding. The real reason it shines is because it handily outperforms that spec. Of course the 5394 also has a flat noise floor out to 100kHz but that would mean nothing if its distortion performance wasn't outstanding. The 5397 proves as much, it too has a flat noise floor out to 100kHz but struggles to give 0.0000x% distortion. I could get reliably get 0.0001% when I tried it, which is similar to what I've seen others manage, but the 5397 also has terrible performance at -1dBfs. The TI is much better in this regard.
If I was using the 5397 and had to pick a non AKM replacement I'd probably go with one of the TI parts too. The TIs aren't bad with regards to noise either. They apparently give a flat noise floor out to around 70kHz, which is really quite good.
Given the QA402 design remit, a quick replacement for the QA401 with the AKM factory fire, I can see why they went with TI parts. They are readily available and with roughly comparable performance to the QA401. In some areas the QA401 was better and in some the QA402 will be better. You'll still be getting a capable piece of equipment for an attractive price.
It's a real shame that TI have decided there isn't any money in premium audio converters. Obviously there's enough because that's all ESS really do. Only recently have they added more cost effective parts. Given the 10+ years since the 1792/4220 I'm sure TI could make some significant improvements.
For cost effective measurement gear the 4490 hit a sweet spot though of being easy to implement and inexpensive.
The interesting thing about the 1792/4 is their digital filter in sharp mode. This is, aside from its latency, still one of the best in any of the new DAC chips. It has 130dB of stop band attenuation, which completely mitigates any aliasing products within the devices signal to noise ratio. It's also got very low passband ripple. Most modern ones come with 120dB or, in the AK4499s case, only 100dB. The AK4499 has been highly criticised on this as aliasing products creep in and affect its performance. Certainly not in any major way, but it's still a point that could be better. If TI could do it almost 20 years ago why can't you?
AKM know about this too and in their AK4498 they've included a filter with 150dB stop band attentuation to completely eliminate aliasing products within the devices signal to noise ratio.
This is a minor point and in most testing scenarios it wouldn't be an issue, it's just not as good as it could be. The 1792/4 won't have these problems though.
Individuals have posted measurements of the top TI ADCs delivering 0.0000x% distortion when the input signal is reduced from maximum, better than spec would indicate. In a way this is similar to the AK5394. If you go by the data sheet AKM provide it's performance is good but not outstanding. The real reason it shines is because it handily outperforms that spec. Of course the 5394 also has a flat noise floor out to 100kHz but that would mean nothing if its distortion performance wasn't outstanding. The 5397 proves as much, it too has a flat noise floor out to 100kHz but struggles to give 0.0000x% distortion. I could get reliably get 0.0001% when I tried it, which is similar to what I've seen others manage, but the 5397 also has terrible performance at -1dBfs. The TI is much better in this regard.
If I was using the 5397 and had to pick a non AKM replacement I'd probably go with one of the TI parts too. The TIs aren't bad with regards to noise either. They apparently give a flat noise floor out to around 70kHz, which is really quite good.
Given the QA402 design remit, a quick replacement for the QA401 with the AKM factory fire, I can see why they went with TI parts. They are readily available and with roughly comparable performance to the QA401. In some areas the QA401 was better and in some the QA402 will be better. You'll still be getting a capable piece of equipment for an attractive price.
It's a real shame that TI have decided there isn't any money in premium audio converters. Obviously there's enough because that's all ESS really do. Only recently have they added more cost effective parts. Given the 10+ years since the 1792/4220 I'm sure TI could make some significant improvements.
Thanks for the summary. I had started wading through the threads but this really made it easier to understand!
Crystal also used to be great but have slowed down new releases to a trickle.
Of course aliasing is bad for excellent measurments. I wonder if it is really bad for listening, since with music, the content varies, so the aliasing does to and is really just inaudible noise. I would tend to think distortion is worse, but then -105 dB distortion is in audible in all likelyhood, too.
Crystal also used to be great but have slowed down new releases to a trickle.
Of course aliasing is bad for excellent measurments. I wonder if it is really bad for listening, since with music, the content varies, so the aliasing does to and is really just inaudible noise. I would tend to think distortion is worse, but then -105 dB distortion is in audible in all likelyhood, too.
Last edited:
There is an interesting discussion in the QA401-thread about the choices for QA402 as well, from here: https://www.diyaudio.com/forums/equ...-quantasylum-qa400-qa401-278.html#post6437799
I had a good meeting with the TI program manager for the audio chips. He explained that the economics aren't there for the chips with SNR much greater than 100 dB. Given the cost of a mask set you really need to sell 1,000,000 to make any money. Low cost 100 dB and you can sell into a phone and then you can justify the investment. TI has many other markets with higher demand for that investment unfortunately. Now if there are other valid markets its easier.
The ESS stuff is pretty amazing business since in the phones its a big piece of the BOM. Qualcomm is going after them which may be a death knell for that chip. ESS was not getting the same $$ from the phone guys it gets from the audio guys but volume talks. Still the life cycle for phones is brutal.
The ESS stuff is pretty amazing business since in the phones its a big piece of the BOM. Qualcomm is going after them which may be a death knell for that chip. ESS was not getting the same $$ from the phone guys it gets from the audio guys but volume talks. Still the life cycle for phones is brutal.
Yep, high-spec analog audio outputs on phones are all but dead, too. The only phones with 3.5mm outputs are/will be budget only. The market is moving to BT or DAC in the cable/dongle if it must be wired. ESS got their chips in some mediocre phones from LG, but I hope they don't bet on sustaining that. I suspect that fashion trend will disappear with the exception of some niche phones no one wants that won't sell.
Maybe they have a shot at the dongle DAC market, though. It's going to be hard to beat Cirrus who already has a few good ICs and is established there.
I guess we should be grateful that TI even makes audio chips still. AD seems to want nothing to do with the audio market. Funny that AD1955 is still pretty good, and well characterized in the datasheet, but who wants to touch an 18 year old chip with a 5V digital supply in a new design.
Maybe they have a shot at the dongle DAC market, though. It's going to be hard to beat Cirrus who already has a few good ICs and is established there.
I guess we should be grateful that TI even makes audio chips still. AD seems to want nothing to do with the audio market. Funny that AD1955 is still pretty good, and well characterized in the datasheet, but who wants to touch an 18 year old chip with a 5V digital supply in a new design.
Last edited:
Trusty arguments but not attractive ! Usually buyers wait for upgrades
Some do, some need something now.
Top of the line AVRs like the old Yamahas (prior to HDMI) had mono DACs, and the top of the line Onkyos had stereo PCM1796. Not sure what the situation is now. With most HDMI AVRs having extremely bad audio processing (as per Amir's measurements), using better DACs would be pure marketing anyway.
You are right, I wait since many months for QA102, and I can wait more times until a perfect fixed analyser. I expect QA102 as high-end product for DIY users not as commercials brands developments (NTI...). Second hand or soundcard are solutions for people in hurry.
In some ways it doesn't matter if it's an improvement over the QA401 or not, since the QA401 is no longer available. Furthermore, can anyone else list something in this price range that is competitive? Next most economical option (on the new market) that I know of is the Averlab, and that's about $3000.
The big improvement in the QA402 (IMO) is that it can handle slightly higher input voltages. That means you can test bigger amplifiers without having to deal with an attenuator. XLR inputs on the front panel is another bonus, since the double-bnc for balanced connections is unusual and annoying.
It'd be nice if it could do the 150 Vrms that some of the bigger analyzers could though, since some of us do occasionally work on really big power amplifiers.
Also, how does the CS5381 fit into the ADC mix?
The big improvement in the QA402 (IMO) is that it can handle slightly higher input voltages. That means you can test bigger amplifiers without having to deal with an attenuator. XLR inputs on the front panel is another bonus, since the double-bnc for balanced connections is unusual and annoying.
It'd be nice if it could do the 150 Vrms that some of the bigger analyzers could though, since some of us do occasionally work on really big power amplifiers.
Also, how does the CS5381 fit into the ADC mix?
They got their 9118 and 9218P into LG G and V series phones. Did they even get their 9219C with integrated MQA rendering into any mass market product?
What kind of mass market product was there that used 9026 / 9028 / 9038, their true top of the line DACs?
Per the last update (QA402 Audio Analyzer – QuantAsylum) of the 8th of May the QA402 should already be shipping normally both inside and outside of the US, however there's no "add to cart" button or indication of how to order... Could someone from QA give an update on the ordering situation? Many thanks,
Many thanks for the info, I'll drop them a line too, see if I have better luck.
Edit: found this blog post from the 3rd of June. Although the QA402 isn't mentioned specifically, the last paragraph points to the whole portfolio still being affected by the silicon shortage. In any case it would be nice if they set up a pre-order or at least some sort of waiting list for those of us interested but not in a hurry...
Last edited: