AT2380v1 // 2 Channels digitally controlled stepped attenuator

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I tried a 12V battery to feed the AT2380.

I remember to you, that the tracking is to the Channel-1

The most noise has gone, remains only the hum of 50Hz and some spikes but in lower levels.

Sig-3dB, Att-3.5dB Channel1 batt.jpg

At the next, I joined the gnd of enclosure and the rear panel on channel 1, now the spikes has gone almost, but not the 50Hz of hum (decreased a little).

Sig-3dB, Att-3.5dB Channel1 batt and screw.jpg

I don't know if it is my fault, but at the Build Instructions Manual at the page 4 says "For proper shielding, the enclosure and the panels must be connected to a signal ground. To avoid ground loop, each attenuator channel have it’s own ground, but only one of them must be connected to the enclosure.

If you want to tracking the Channel-1, according to this instructions (see the Shield Table at the page4) you must join only the Channel-2 as shield and not this of Channel-1
But, I see if you joined and the Channel-1 the spikes eliminated!

Frex, I think to investigate what is happening with 50Hz hum. This appears either battery or external PSU and at the tracking channel not to the other.
I think to remove the D1 diode at the input and jumper this.
 
Last edited:
Hello Lemon,

I'm very busy until next week-end, so then I will look also on my side.
I don't remember all i do in my measurement setup.

Did you have tried to put the AT2380 enclosure far from the top you desk ?
(By putting an isolated box between, for example).

Do you have a source near the setup that radiated 50Hz field ?
(Like a main transformer).

Frex
 
Hi Frex,
Take your time, we have a lot time for our love (diy).
For your questions, yes I have tried many things but they didn't work. No, there is no any strong source of 50Hz that radiates the equipment.
All my others equipment that it is close to AT....no problem with 50Hz.

Don't forget that the other channel, the Channel-2 that isn't tracking hasn't any 50Hz, it is very clean (post 97).
In the meanwhile, I am going to done some other mods (like D1) and update here with my results
 
I searched a lot the 50Hz hum.
There is no any cause from on board reg or other equipment of supply. I bypassed the LT317 and feed outside the 5V and there is no any improvement.

Finally, in my situation the solution was done with a connection of gnd of digital pcb with the Channel-1 shield (see the follow picture the yellow cable).

solution.JPG

With this solution, the tracking channel-1 works fine without hum or spikes, like the Channel-2 with no tracking. This capture was done with Rigol PSU.

final.jpg

Now, it's time to concentrate with the calibration and how the auto tracking works because is somewhat difficult but I will wait and other members like hallodeletue and bk856er.
 
Did you include auto-tracking? Did the calibration work? I am struggling quite a bit with mine, the 2 trimmers aren't linear, but jump around quite a bit... Hence I never get both adjustments right.

Best regards,

hallodeletue

I did a precisely calibration and I make the target 0.00/-80.dBV with the two P1/P2 trimmers.
You must not turn a lot of one trimmer, but a little each trimmer at interval time of two leds.
Now, auto-tracking works fine and the dBV level also.

I must to say, that I followed the opposite series at interval time of two leds. Otherwise, I had an opposite result!
I don't know why.
 
AT2380 Firmware - update v0.25

Hello,


After some users feedback, i made a new CPLD update on the AT2380v2 attenuator.
In fact, when the attenuator settings are changed, because the relays switch on/off time
are not symmetric and then can produce a large output level between these time gap.

With the current firmware v0.24, you'll get output like this :
(Note: AT2380 input is 1kHz sine wave 1 V rms,
Jump setting between -64 dB to -63.5 dB - worst case ).

V0.24_64_to_63.5.png


As we can see, for about 1ms the output level is near input and there is almost no attenuation
until relays final state are reached.
This spike cause audible click and it's not safe for for output load since level can be not
as expected with selected attenuation.

In the new updated firmware v0.25, i added some delay when relays are switched
"on to off" but not when "off to on".
This allow to completely remove previous behavior as show scope screen-shoot below :

V0.25_64_to_63.5.png



We can see now that with exactly same input level and setting (jump from -64 to -65 dB)
that we can zoom vertical scale much and see a zero level when relay "break before make" operate.
So, there is no spikes anymore...


The update firmware v0.25 can be downloaded HERE (pof file only).
I continue to work a little on the design and the complete CPLD source will
be provided on Github soon for easiest update and project monitoring.
Regards.

Frex
 
Please let me know why I might be wrong, but what would be the problems with using a digital potentiometer chip like Analog Devices AD5124? It has 4 channels, which could presumably be put into two parallel pairs to halve the 8% tolerance to 4% if you were happy to have 50k or 5k total resistance.
Easily controlled with an Arduino and most have non-volatile memory, too.

Any thoughts?

All the best,
Dax.
 
Bonjour Frex! Thanks for the quick reply!

Excellent point. I had a look through the ICs available and there is an AD5291/5291 (256 vs 1024 steps) which seems to have a THD figure -103dB for the 100k version. The 50k is -101dB and the 20k is a comparatively ordinary -93dB. These are all 1-channel units as the THD figures go up to the levels you mentioned for the part I originally quoted.

I checked some commercial monitor controllers, but their THD specs were in % rather than dB and I couldn't find any resources on how to convert (or even if they're analogous at all).

https://www.analog.com/media/en/technical-documentation/data-sheets/AD5291_5292.pdf
 
Member
Joined 2004
Paid Member
There is more than a little smoke and mirrors in the Benchmark writeup. The story about pots generating distortion at the contacts is only true for the cheesiest grade. The distortion of a premium Alps, Noble or Bourns pot is well below -150 dB (I have measured them). And you don't need that many relays for an active attenuator, plus the range below -60 dB is not useful. here is a much more efficient design: Audio Volume Relay Attenuator with IR Control. I build a variation many years ago using Vishay S102 resistors. The story about resistors generating distortion at audio frequencies is very specific and true for Vishay foil resistors at 20 Hz and below running near rated power. And the distortion is still very low.

As for chip volume controls this is quite good: https://www.njr.com/electronic_device/PDF/MUSES72320_E.pdf. The .001% translates into -100 dB, .0005% is -106 dB. Mostly contribution from the associated opamp.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.