• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Vintage Audio: Measurements of Arcam Black Box 3 mk1

LTig

Addicted to Fun and Learning
Joined
Feb 27, 2019
Messages
677
Likes
822
Location
Europe
#1
As requested by @andreasmaaan here I made some measurements of an old DAC: the Arcam Black Box 3 MK1 which I bought new in 1991 (without MK1, see below). It retailed for around € 780 then (converted). Here are the front and the back:

arcam bb3 front.jpg

arcam bb3 back.jpg
It supports 44.1 and 48 kHz samplerate with 16 bits - that was the standard back then ...

Somewhere later around 1997 I fell victim to op-amp rolling and replaced the original op-amps (LM627CN) by OPA134PA. Two can be seen on this photo of the internals on the right PCB (audio mother board).The LF411 is used as DC servo. Two others are hidden on the backside of the left PCB (digital audio board). The result is my version MK1:

arcam bb3 intern.jpg


The measurements are done with REW, using an RME ADI-2 PRO fs, an old Edirol UA-25 and a brand new SIGLENT SDS 1202X-E (2 channel 200 MHz digital scope).

Lets have a look to the output level (measured with the digital scope):

arcam bb3 1kHz sinus 0 dBFS.jpg

The left channel (yellow) delivers 2.50 Vrms, the right channel 2.35 Vrms, a difference of 0.5 dB.

For the THD, IMD and Jitter measurements I used the RME only, feeding the Toslink input of the BB3 with 44/16 sinus signals.

THD 1kHz 0dBFS 32k x1.jpg


32 times averaging reveals remains of the power supply:

THD 1kHz 0dBFS 32k x32.jpg


This is not 16 bit performance.

IMD CCIF -3.1 dBFS 32k x1.jpg

The actual IMD is not that bad, but there is a lot of spurious stuff. With 32 times averaging we can see that it is not noise:

IMD CCIF -3.1 dBFS 32k x32.jpg

EDIT: while repeating the measurements with the fixed power supply and back in original state I realized that I did not enable dither during the IMD measurements. With dither enabled the result is much better: the spikes are still there but reduced in amplitude by about 10 dB.

Jitter suppression looks really awful (using J-Test signal):

J-Test 32k x32.jpg


And this stuff covers the full spectrum:

J-Test 32k x32 full.jpg


At last I measured the frequency response. Here I used the Edirol to feed the BB3 with 44/16 signals and used the RME with 384/24 to cover a wide band spectrum. REW did not allow me to display the full spectrum (the RME meausres until 180 kHz) so I can show only up to 100 kHz. First white noise with heavy averaging:

Frequency response white noise 256k x338.jpg


Oh oh - this does not like like a good reconstruction filter. I would expect a lot of aliasing here, so let's check with a 19.1 kHz sinus (same as JA uses in his tests for stereophile), 64 averages:
Antialiasing 19.1 kHz 0 dBFS 256k x64.jpg


The aliased image of the 19.1kHz tone at 0 dBFS is suppressed by 72 dB. This is much more than the white noise result suggests (beats my why) but it should be at less than -90 dB.

Conclusion

We should be glad that we live in modern times where 100 € buy you a DAC which measures much better than this vintage gear. I would not say that the BB3 was high-end (that would have been Proceed or Wadia which were at least twice as expensive), but it was not cheap either.

Please remember that this unit is not in the state as it left production (it's MK1!). Without reinserting the original op-amps and a retest we cannot be sure that the THD and IMD measurements reflect the original performance. The jitter and aliasing problems however should not be affected - but we never know ...

EDIT: corrected the type of the original op-amps (from LM327 to LM627CN).
 

Attachments

Last edited:

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
2,524
Likes
4,162
Location
Gold Coast, Queensland, Australia
#3
Nice post. Well done. :)

What load did you have when measuring the output level BTW? Sometimes, level differences can show up when the outputs are unloaded and even 10K can bring them back in line.

Interesting to see the old SAA-7321- the first Bitstream low cost dual channel chip being used one per channel in differential mode as Bob Stuart's Meridian 203 did also around the same time.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
6,685
Likes
6,699
#5
Yep very nice post. Good job on the testing.

About the white noise vs 19.1 khz imaging result, it is because of the nature of white noise. A full on white noise signal will display lower due to FFT gain. At 256k FFT over 176 khz bandwidth (of the RME) it is about 42 db lower in level than 0. So if you use it as your reference the imaging in the ultrasonic region is the difference shown plus about 42 db with noise. So you get something in the 70's I'd say with your graph. Closer to the 72 db you get using 19.1 khz. So it does all make sense.
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
2,524
Likes
4,162
Location
Gold Coast, Queensland, Australia
#8
and a brand new SIGLENT SDS 1202X-E
Please post some impressions of the new DSO, it's a worthy challenger to the venerable Rigol DS1054Z in many ways and arguably better in some parameters.

:)
 

sergeauckland

Addicted to Fun and Learning
Patreon Donor
Joined
Mar 16, 2016
Messages
960
Likes
1,762
Location
Suffolk UK
#9
Very interesting and thanks for making and posting the measurements. What interests me about this is that whilst undoubtedly modern DACs, even quite cheap ones, measure much better, I've not seen any blind listening tests between a good ~30 year old DAC and something modern that would indicate that the old DAC is sonically inferior. Have there been any such tests published?

I still use a Meridian 206 CD player, a Technics SL-PG590 and a Squeezebox Touch, all with their own internal DACs, and have failed to hear any difference whatsoever when level matched.

S.
 
Joined
Apr 7, 2018
Messages
29
Likes
14
#11
Wow! I used to own one of these back in the days before family commitments curbed my spending. Nice to see my ASR recommended £60 DAC blows it out of the water. :)

Can't believe the 90's are viewed as vintage already though. I'll except no argument you made a mistake there. :D
 
Last edited:

sergeauckland

Addicted to Fun and Learning
Patreon Donor
Joined
Mar 16, 2016
Messages
960
Likes
1,762
Location
Suffolk UK
#12
Wow! I used to own one of these back in the days before family commitments curbed my spending. Nice to see my ASR recommended £60 DAC blows it out of the water. :)

Can't believe the 90's are viewed as vintage already though. I'll except no argument you made a mistake there. :D
I struggle to see anything after 1980 as 'vintage'. Vintage has tubes in it, or at a push, geranium transistors (and zenobia diodes)!

S
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
2,524
Likes
4,162
Location
Gold Coast, Queensland, Australia
#13
I struggle to see anything after 1980 as 'vintage'.
Back in the the early 90s, we were calling 60s and 70s gear, vintage, so it makes sense that nearly 30 years later, 80s and 90s gear is now officially vintage too. :)
 

LTig

Addicted to Fun and Learning
Joined
Feb 27, 2019
Messages
677
Likes
822
Location
Europe
#14
Yep very nice post. Good job on the testing.

About the white noise vs 19.1 khz imaging result, it is because of the nature of white noise. A full on white noise signal will display lower due to FFT gain. At 256k FFT over 176 khz bandwidth (of the RME) it is about 42 db lower in level than 0. So if you use it as your reference the imaging in the ultrasonic region is the difference shown plus about 42 db with noise. So you get something in the 70's I'd say with your graph. Closer to the 72 db you get using 19.1 khz. So it does all make sense.
The 42 dB shift makes sense, but why would it apply to the pass band only and not also to the stop band?
 

LTig

Addicted to Fun and Learning
Joined
Feb 27, 2019
Messages
677
Likes
822
Location
Europe
#16
Please post some impressions of the new DSO, it's a worthy challenger to the venerable Rigol DS1054Z in many ways and arguably better in some parameters.

:)
Will do, but first I have to become familiar with it. This is my first digital scope (it replaces an old Tek 465, 2ch 100MHz) and there is only a quick setup manual. I guess I have to download the full manual and study it. One thing I can say already: the display feels like an analog scope with a long fade out time (when enabled). I once had a very old HP 130 from 1965, 2ch 500 kHz (yes, kiloHz), with a fade out time of 40 s!

I have very little experience with digital scopes yet. In the office we have a Tek 2465 (4ch 300 MHz) and an early digital scope (4ch 500 MHz), but in those few situations where I needed a scope I always used the 2465 because I know how to set it up and just concluded that it will be faster this way than starting to read the manual of the digital scope. My younger colleagues prefer the digital scope though, because if its default button :facepalm:
 

LTig

Addicted to Fun and Learning
Joined
Feb 27, 2019
Messages
677
Likes
822
Location
Europe
#17
Can't believe the 90's are viewed as vintage already though. I'll except no argument you made a mistake there. :D
I see it as vintage digital audio gear. CD came in 1983, the first consumer DACs AFAIK a few years late. Seen from this perspective a DAC made in 1991 is really old (28 years), ergo vintage.
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
2,524
Likes
4,162
Location
Gold Coast, Queensland, Australia
#18
where I needed a scope I always used the 2465 because I know how to set it up and just concluded that it will be faster this way than starting to read the manual of the digital scope.
Same here. My analog scope (vintage Trio 60MHz) is still my go-to scope. I can operate it virtually blindfolded as I've owned it and a few others for many decades and they never let me down.

The DSO is still a learning curve several years on. They make you lazy too- all the math is done in the scope. That said, they are amazing tools.
 
Joined
Feb 9, 2018
Messages
151
Likes
80
#19
It's just practice...

I spent months in front of an HP8568B when developing RF synthesis technology for the early days of Vodafone. My fingers used to fly around the keys to take measurements, without thinking. 30-odd years on, I'd struggle to set the centre frequency...
 

LTig

Addicted to Fun and Learning
Joined
Feb 27, 2019
Messages
677
Likes
822
Location
Europe
#20
What load did you have when measuring the output level BTW? Sometimes, level differences can show up when the outputs are unloaded and even 10K can bring them back in line.
Just the scope, that is 1 MOhm. Looking at the schematics I cannot see that a lower resistance would change the levels.

I just bought a metal box and some stuff (0.1% resistors) to build a multi purpose adapter which allows me to change the load or add attenuators. When it's done I can test with different loads.
 
Top Bottom