Unbranded ADC from eBay - measurements

[dominikz]

Recently I was looking for a cheap ADC and found a generic, unbranded one on eBay for just under 10$ (UPC: 6927373037416). Here's a few pictures of the device:

My plan was to use an ADC to connect my turntable to a digital input of miniDSP OpenDRC-DI, but I didn't really want to spend much on this and just hoped the cheapest one I find would be satisfactory

The measurements you'll see next were done with the RME Babyface soundcard, first generation - measured here. The ADC was connected to optical SPDIF input of RME Babyface, clock source was the optical input (i.e. ADC) and the sample rate was 48kHz (native sample rate of ADC). Unbalanced/headphone output of RME babyface was used as the signal generator. The performance of the ADC varied significantly based on the type of PSU used, and in these tests I used a powerbank to power the ADC and get the cleanest response. If using a 'dirtier' PSU the noise would increase by 5-6 dB and dynamic range would go down by a similar amount.

Let's start with the basics, RMAA measurement summary (input level set to recommended -1 dBFS by reducing RME output level by ~7dB):

As we can see, performance was not very good - and this is the best I was able to get.

Here's the frequency response:

As we can see the LF -3 dB point is already at 55-60 Hz (not the same for the two channels), and the -6 dB point is in the mid thirties - quite disappointing. Good news is that there is no visible ringing in the HF, the response looks pretty smooth there.

Next let's see the spectrum of a 1 kHz sine at -1 dBFS peak:

As we can see, there's a lot of noise and non-harmonic tones in it, and we can even see pretty high level IMD products around the test 1kHz tone. THD is also pretty high at this level, dominated by the 2nd harmonic.

Let's look now at the THD vs level sweep at 1kHz:

We can see that THD minimum is achieved at only ~25mV input (0,006-0,01%) and the ADC starts to clip around 0,8V. So not a lot of headroom, though I guess it should be OK to digitize a phono line output (mine is specified at 200 mV nominal at 1 kHz).

Lastly let's look at the THD vs frequency at two input levels, first -17 dBFS RMS (~0,1 Vrms):

And then at -5 dBFS RMS (~0,4 Vrms):

As we can see, distortion is mostly dominated by the 3rd harmonic at the lowest frequencies (which doesn't seem to be modulated by level), and then is taken over by the 2nd harmonic, which increases as input level increases. Otherwise there does not seem to be much frequency dependency of distortion (total THD level seems pretty constant above ~200 Hz).

All in all not a very impressive showing, but on the flip-side it is a pretty cheap device, so it could probably come in handy in non-critical uses.

In my case I decided not to use it with my turntable - after I plugged it in I felt the added distortion and lack of bass was audible - though it's hard for me to say if it was just bias from seeing the measurements first as I'm pretty sure even this performance is probably quite a lot better than what my turntable can achieve. Anyway, for my peace of (OCD-ridden) mind I kept this ADC out of the system

EDIT [2021-06-05]: Added a comparison to the RME ADC (measurements and clips) in post #8.

 

[sergeauckland]

I bought the DAC version, which cost me £3.45 including free postage!
At first, I thought the performance was terrible, but I realised that the DAC had no output capacitors, so had an output DC offset, which upset my measuring kit. Two capacitors later and the measurements were much improved, easily audibly transparent. At £3.45 including postage, just the metal can was worth that much.

It might be worth checking if there's anything simple that'll improve the ADC performance. One might be pleasantly surprised.

S.

 

[amirm]

I tested the DAC-only version: https://www.audiosciencereview.com/forum/index.php?threads/worlds-cheapest-dac-review.16594/

 

[amirm]

I promoted the thread to home page.

 

[GWolfman]

Thanks for contributing.

 

[peanuts]

rolloff starting around 200hz? thats pretty bad regardless of price...

 

[rodrigaj]

I do hope that ADC's become a regular tested item. Lots of turntables out there, not all of which are owned by hipsters, looking inexpensive solutions to digitizing favorite records.

 

[dominikz]

I promoted the thread to home page.

Wow, I'm honored - thanks!

At first, I thought the performance was terrible, but I realised that the DAC had no output capacitors, so a 2v output DC offset, which upset my measuring kit. Two capacitors later and the measurements were much improved, easily audibly transparent. At £3.45 including postage, just the metal can was worth that much.

It might be worth checking if there's anything simple that'll improve the ADC performance. One might be pleasantly surprised.

That's really interesting! Perhaps I'll look into it if I find some time, thanks for the tip!

On a side note, I thought it might be interesting to record an analogue loopback audio clip through this ADC and the ADC in my RME soundcard, and to compare that also to the original clip (pure digital export without passing DAC/ADC). So that's what I did

First a comparison of the two ADCs using the same output set to the same level (~0,7Vrms), RMAA summary:

As we can see, RME ADC is quite a lot better at all points (as expected )

The comparison clips have been recorded to avoid clipping, but in both cases with the same output level, they are loudness matched (using integrated LUFS scale) and can be downloaded here. The song is one of my band's since that's what I have distribution rights for (and also, why not self-promote? ), full song available here.
The clips can be used in e.g. foobar ABX plugin for blinded comparisons. Here's mine:

foo_abx 2.0.6d report
foobar2000 v1.6.2
2021-06-05 22:44:41

File A: Chain the Sun - BtR - eBay ADC loopback.flac
SHA1: 3dff4fa889152e33a9309be0bc815d6d8b6d318e
Gain adjustment: -4.66 dB
File B: Chain the Sun - BtR - RME ADC loopback.flac
SHA1: 6ee80119abfd3726130e2a7be87a18d28e9ef717
Gain adjustment: -4.62 dB

Output:
Default : Primary Sound Driver
Crossfading: NO

22:44:41 : Test started.
22:46:05 : 01/01
22:46:20 : 02/02
22:46:38 : 03/03
22:46:51 : 04/04
22:47:01 : 05/05
22:47:13 : 06/06
22:47:31 : 07/07
22:47:47 : 08/08
22:47:54 : 09/09
22:48:19 : 10/10
22:48:31 : 11/11
22:48:39 : 12/12
22:48:39 : Test finished.

 ----------
Total: 12/12
p-value: 0.0002 (0.02%)

 -- signature --
f8b9a120470157aac852349a7b3595e22b82220d

I have to say I was surprised that the audible difference was smaller that I thought originally (especially considering the relatively extreme difference in performance between the two), though apparently still possible for me to reliably differentiate - so yay!
If someone else runs the trials I'd be very interested to see more results!

 

[Blumlein 88]

I do hope that ADC's become a regular tested item. Lots of turntables out there, not all of which are owned by hipsters, looking inexpensive solutions to digitizing favorite records.

You can pretty safely use something like a Focusrite 2i2 or similar items. Even 2nd hand 1st and 2nd gen versions will do just fine. These aren't state of the art, but comfortably eclipse the performance envelope of turntables. So you'll have pretty much no degradation digitizing favorite LP's.

Also Motu M2 once it become back in stock regularly is a good choice.

 

[Robbo99999]

Recently I was looking for a cheap ADC and found a generic, unbranded one on eBay for just under 10$ (UPC: 6927373037416). Here's a few pictures of the device:
View attachment 133907View attachment 133908
View attachment 133909

My plan was to use an ADC to connect my turntable to a digital input of miniDSP OpenDRC-DI, but I didn't really want to spend much on this and just hoped the cheapest one I find would be satisfactory

The measurements you'll see next were done with the RME Babyface soundcard, first generation - measured here. The ADC was connected to optical SPDIF input of RME Babyface, clock source was the optical input (i.e. ADC) and the sample rate was 48kHz (native sample rate of ADC). Unbalanced/headphone output of RME babyface was used as the signal generator. The performance of the ADC varied significantly based on the type of PSU used, and in these tests I used a powerbank to power the ADC and get the cleanest response. If using a 'dirtier' PSU the noise would increase by 5-6 dB and dynamic range would go down by a similar amount.

Let's start with the basics, RMAA measurement summary (input level set to recommended -1 dBFS by reducing RME output level by ~7dB):
View attachment 133910
As we can see, performance was not very good - and this is the best I was able to get.

Here's the frequency response:
View attachment 133911
As we can see the LF -3 dB point is already at 55-60 Hz (not the same for the two channels), and the -6 dB point is in the mid thirties - quite disappointing. Good news is that there is no visible ringing in the HF, the response looks pretty smooth there.

Next let's see the spectrum of a 1 kHz sine at -1 dBFS peak:
View attachment 133913
As we can see, there's a lot of noise and non-harmonic tones in it, and we can even see pretty high level IMD products around the test 1kHz tone. THD is also pretty high at this level, dominated by the 2nd harmonic.

Let's look now at the THD vs level sweep at 1kHz:
View attachment 133919
We can see that THD minimum is achieved at only ~25mV input (0,006-0,01%) and the ADC starts to clip around 0,8V. So not a lot of headroom, though I guess it should be OK to digitize a phono line output (mine is specified at 200 mV nominal at 1 kHz).

Lastly let's look at the THD vs frequency at two input levels, first -17 dBFS RMS (~0,1 Vrms):
View attachment 133921
And then at -5 dBFS RMS (~0,4 Vrms):
View attachment 133923
As we can see, distortion is mostly dominated by the 3rd harmonic at the lowest frequencies (which doesn't seem to be modulated by level), and then is taken over by the 2nd harmonic, which increases as input level increases. Otherwise there does not seem to be much frequency dependency of distortion (total THD level seems pretty constant above ~200 Hz).

All in all not a very impressive showing, but on the flip-side it is a pretty cheap device, so it could probably come in handy in non-critical uses.

In my case I decided not to use it with my turntable - after I plugged it in I felt the added distortion and lack of bass was audible - though it's hard for me to say if it was just bias from seeing the measurements first as I'm pretty sure even this performance is probably quite a lot better than what my turntable can achieve. Anyway, for my peace of (OCD-ridden) mind I kept this ADC out of the system

EDIT [2021-06-05]: Added a comparison to the RME ADC (measurements and clips) in post #8.

I don't understand, why does the bass have to taper off at 100Hz, to me that's not covering the basics of the frequency spectrum, why's it doing that?

 

[phoenixsong]

You can pretty safely use something like a Focusrite 2i2 or similar items. Even 2nd hand 1st and 2nd gen versions will do just fine. These aren't state of the art, but comfortably eclipse the performance envelope of turntables. So you'll have pretty much no degradation digitizing favorite LP's.

Also Motu M2 once it become back in stock regularly is a good choice.

I'm also curious about Radial's offerings. Many of these pro audio companies seem to have decent products

 

[phoenixdogfan]

Recently I was looking for a cheap ADC and found a generic, unbranded one on eBay for just under 10$ (UPC: 6927373037416). Here's a few pictures of the device:
View attachment 133907View attachment 133908
View attachment 133909

My plan was to use an ADC to connect my turntable to a digital input of miniDSP OpenDRC-DI, but I didn't really want to spend much on this and just hoped the cheapest one I find would be satisfactory

The measurements you'll see next were done with the RME Babyface soundcard, first generation - measured here. The ADC was connected to optical SPDIF input of RME Babyface, clock source was the optical input (i.e. ADC) and the sample rate was 48kHz (native sample rate of ADC). Unbalanced/headphone output of RME babyface was used as the signal generator. The performance of the ADC varied significantly based on the type of PSU used, and in these tests I used a powerbank to power the ADC and get the cleanest response. If using a 'dirtier' PSU the noise would increase by 5-6 dB and dynamic range would go down by a similar amount.

Let's start with the basics, RMAA measurement summary (input level set to recommended -1 dBFS by reducing RME output level by ~7dB):
View attachment 133910
As we can see, performance was not very good - and this is the best I was able to get.

Here's the frequency response:
View attachment 133911
As we can see the LF -3 dB point is already at 55-60 Hz (not the same for the two channels), and the -6 dB point is in the mid thirties - quite disappointing. Good news is that there is no visible ringing in the HF, the response looks pretty smooth there.

Next let's see the spectrum of a 1 kHz sine at -1 dBFS peak:
View attachment 133913
As we can see, there's a lot of noise and non-harmonic tones in it, and we can even see pretty high level IMD products around the test 1kHz tone. THD is also pretty high at this level, dominated by the 2nd harmonic.

Let's look now at the THD vs level sweep at 1kHz:
View attachment 133919
We can see that THD minimum is achieved at only ~25mV input (0,006-0,01%) and the ADC starts to clip around 0,8V. So not a lot of headroom, though I guess it should be OK to digitize a phono line output (mine is specified at 200 mV nominal at 1 kHz).

Lastly let's look at the THD vs frequency at two input levels, first -17 dBFS RMS (~0,1 Vrms):
View attachment 133921
And then at -5 dBFS RMS (~0,4 Vrms):
View attachment 133923
As we can see, distortion is mostly dominated by the 3rd harmonic at the lowest frequencies (which doesn't seem to be modulated by level), and then is taken over by the 2nd harmonic, which increases as input level increases. Otherwise there does not seem to be much frequency dependency of distortion (total THD level seems pretty constant above ~200 Hz).

All in all not a very impressive showing, but on the flip-side it is a pretty cheap device, so it could probably come in handy in non-critical uses.

In my case I decided not to use it with my turntable - after I plugged it in I felt the added distortion and lack of bass was audible - though it's hard for me to say if it was just bias from seeing the measurements first as I'm pretty sure even this performance is probably quite a lot better than what my turntable can achieve. Anyway, for my peace of (OCD-ridden) mind I kept this ADC out of the system

EDIT [2021-06-05]: Added a comparison to the RME ADC (measurements and clips) in post #8.

I would have thought for analog a 56 db Sinad would have been good enough.

 

[Blumlein 88]

I'm also curious about Radial's offerings. Many of these pro audio companies seem to have decent products

They have various boxes often switching for DI or microphones. I'm not aware of them making a recording interface (but maybe they do).

 

[phoenixsong]

They have various boxes often switching for DI or microphones. I'm not aware of them making a recording interface (but maybe they do).

Presenter - Radial Engineering something like this?

 

[Blumlein 88]

Presenter - Radial Engineering something like this?

That could be used. Usually mixers only allow 16 bit recording. I don't know in this case as they don't say. Recordings records which may have tics and pops a bit of extra headroom is a good idea.

 

[xema]

Good enough at this price, as long as it can work.

 

[McFly]

Timely review. Thanks for this. Anyone have a better suggestion for RCA analogue to SPDIF?

 

[restorer-john]

I've got one the same here someplace too, will see how it compares. I remember looking inside and then throwing it in the parts bin...

 

[dominikz]

I would have thought for analog a 56 db Sinad would have been good enough.

In principle I agree with you. As I said, in my case it might just be bias from seeing the measurements. But I can also imagine potential issues with gain staging, given that the device has low headroom on one side and relatively low THD+N on the other. There's also the early bass roll-off that is perfectly audible (I was able to reliably ABX it when level-matched, see post #8) - for me this was the main issue.
In summary, I believe this ADC probably can work OK as an interface to digitize a turntable output, but some care would be needed to set an appropriate input level and possibly also to counteract some of the bass roll-off. Personally I'd want something that would give me more peace of mind
However, given the price IMHO it is hard to complain too much about the measured performance

 

[sergeauckland]

I wonder whether the poor LF performance is due to small input coupling capacitors, assuming it has some! It seems odd to me for an ADC chip, of whatever cost, to be so poor a low frequencies. If so, then it's a cheap and quick fix.

S.