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.
Hello all,

Here is my new little DIYproject, the "AT2380".

As many DIYers here, i often needed to attenuate a signal to fit better with the input level of the target system.
This is a tool that i need to perform some tests on my other design, the AA2380 ADC.

So, i was thinking to building an attenuator for a long time.
There is of course some already built, but good attenuators are generally dedicated for RF application with 50/75 Ohms impedance.
This is a too low impedance for the main audio and instrumentation applications that i target.

I work on this project for some months now, in parallel with some others.
I designed it to meet my needs and to be a cost effective solution to share it with all interested DIYers.


The main specifications of the attenuator are :


  • full digital control (5M570 CPLD, VHDL coded).
  • 2 pseudo isolated attenuators Channels
  • Zin= 600 Ohms , Zout = ~ 0 to 300 Ohms (customizable)
  • Flat frequency response (± 0,05dB) from at least DC..1MHz (± 3dB to 10 MHz).
  • 0 to 127,5 dB attenuation in 0,5 dB step (8 bits / 256 steps)
  • Direct dB attenuation setting displayed on 4 digits seven segments LEDs.
  • Five selectable dB steps : 0.5 ,1 , 3, 6 and 10 dB by encoder increment.
  • Optional output level auto-tracking mode from +10 dBV to – 85 dBV (attenuation only) .
  • Full internal clock enable switch, for ultra quiet operation mode.
  • Calibration mode for dBV measurement
  • Relay fast refresh mode for audio volume control application
  • Bluetooth UART link to control attenuator and measure output dBV level
  • Shielded design to improve RF performance and noise immunity .
  • Low cost design and easy to find parts.
  • 100x160mm PCB size, fit in Hammond 1455 series extruded aluminum enclosure.
  • Front and rear panel made in PCB will all holes drilled and with gold silkscreen.
  • Through holes axial or SMD (1206) ladder attenuator resistors types.


These specs are preliminary as i don't have yet tested it, but it's the target.
Each attenuator channel use a ladder resistors design switched by 8 relays.
So, with 0.5dB resolution the attenuation can be controlled over 0 to -127.5 dB range.
(You can see how work a ladder log attenuator and calculate value here.)

The control of the attenuation is done by a rotary encoder with push button.
For easiest setting, the dB step by encoder increment can be choose between 0.5-1-3-6 or 10 dB.
The actual attenuation level is show on a 7 segments led display.

There is also the possibility to use it in tracking level mode.
When this mode is enabled, the output level of the attenuator track the desired level needed in the range of +10dBV to -90dBV (3Vrms to 30µVrms).

At this time, the prototype PCB has been send for manufacturing and i will have it in next week.
I will post a picture as soon it is in my hands.

You can show below the synoptic of the design :

AT2380v2.1_synoptic.png


Others infos will follow soon, and i hope that the project will interest others DIYers.
Regards.


Update of the 20/11/2017
------------------------------------


FInal V2.0 released
The full schematic and the manual of the attenuator is available below :

Full schematics : AT2380v2_sch.pdf

Operating manual and specifications : AT2380V1_Manual_v0.pdf

Functional Synoptic :
AT2380v2 functional synoptic

Complete BOM file for both track and not rack, with Mouser price list :
AT2380v2 full bill of material with Mouser price list


I finalize the Gerbers files, and the group-buy for bare PCB will be open on next week-end.
More infos soon. :)




Update of the 26/11/2017 -- PCB GroupBuy started
---------------------------------------------------------------------------


The PCB group buy is started and can be found here :
AT2380v2 2Ch Digitally controlled Stepped Attenuator




Update of the 30/01/2018 -- Firmware Update. (v0.23)
---------------------------------------------------------------------------


I released today the new 0.23 CPLD firmware that solve some bugs and add new features :
- Version display at startup
- Calibration mode.
- Relays fast mode operation (for audio volume control applications)
- Bluetooth UART mode attenuator control

Because of these many modifications, i have made a new design folder (v2.1),
provided to all PCB owners.

Some files are available to download on the links below :

AT2380_SynopticV2.1
AT2380v2.1_schematics
AT2380_Manualv1.1
AT2380V2_Bill_of_material_Rev1



Note that bare PCB are available, if interested send me a PM.


Frex
 
Last edited:
Hello,

Thank you Jcga, you are faithful to my projects :)

I received the prototype PCB's, and to show little more
what did it look like, some pictures of them below :


The complete PCB with main board and the attached front and rear panel :
AT2380v1_bare1.JPG


The PCB detached :
AT2380v1_bare2.JPG


A small mechanical verification with front panel :
AT2380v1_bare4.JPG


Same with rear panel :
AT2380v1_bare3.JPG




All seem to fit well for now, but it's just a fast test.
I must now order few missing parts, i will have all very soon.
So next step will be to solder all of that.... :)

I will update soon.
Regards.

Frex
 
Member
Joined 2004
Paid Member
I'm interested too. Adding a USB interface could enhance its utility.
I would not expect the settings below -100 dB to be too accurate. At that level of attenuation stray coupling will dominate. The old GR's would you two sections for the higher attenuations and even then they have issues. At 50 Ohms you can get to higher frequencies but its still an issue. At RF its almost hopeless. On the HP Network analyzers you can put closed terminations on them and it still gets through all the stainless steel.

Here is how GR built their attenuator: http://i670.photobucket.com/albums/vv62/mikeinkcmo/testgear/GR/1450 Step Atten/GR14504.jpg It shielded isolated elements with feedthroughs to the next stage. They get 200 KHz out of it. Those resistors are great but virtually unobtainable. The switch contacts are also really premium.
 
Hello,

I worked a little this week on it and i made the 2 copper shield panels to separate
each attenuator channel. I also cut the red window for the Led display.
Missing parts are ordered, i will have all to be soldered next week.

Demian, I don't have planned to provide USB link, nevertheless it is doable.
There is some unused IO available on en header connector (test purpose)
that can be used with an ftdi chip for example to add usb control.
So, some code will have to be added of course and CPLD is not so big.

About the attenuation obtainable, as soon prototype build,
i would made extensive tests to verify attenuation performance versus frequency
at different attenuation levels. Defining full and clear specifications of
any electronics instrument is essential.

Frex
 
Member
Joined 2004
Paid Member
You may need a shield can/copper around the -60 dB and possibly the -30 dB steps. Those are the ones where parasitics can dominate the attenuation.

Accurate measurement of attenuation at high frequencies is really difficult. There are many possible error mechanisms that make the process difficult. However once done it should be very stable. Also 10V -120 dB is 10 uV. Few instruments can measure that without noise dominating. Usually frequency selective meters are needed. You will need to plan for those challenges.

Lets see how it performs before getting into automation. However that plus a programmable oscillator would be ideal for checking an RIAA curve for example.
 
Hello,

Thank you for your comment Mediatechnology.
Any I/O of the CPLD can be used for a SPI link, and there is a 8 IO free connector
available on the PCB. I add it for test purpose.
So, adding SPI link to control the attenuation level could be done, just need to write
some code with taking in account available cells...

Demian,
As maybe that you can see on the PCB pictures, there is some (4) slot inside PCB.
These slot are intended to add 2 vertical copper shields made in PCB.
After insertion, they are soldered all along the PCB on the "golden line".
It's intended to reduce significantly HF impedance of ground planes ans also to shield channels.
I will post a picture with them inserted this week.

I don't have made much job on the design these day, DIY work in the house
is taking all my time for now...

Frex
 
Parts soldered ! :eek:)

Hello,

I mounted all parts these days. All fit well.
Below, some pictures of the result.

Demian, you can see the two vertical copper shields between
each attenuator channel (not yet soldered).

AT2380_M03.jpg


AT2380_M05.jpg


AT2380_M11.jpg




It has not been yet powered.Time to test now...


Frex
 
Hello,

I worked a lot this week on the CPLD software and now manual control is done and work very well. Output dBV measurement work also fine from -80 to +10dBV.
The last unterminated function is the control loop ofr tracking mode that is not yet completely functional. That will come soon, and then i will do a little video to present the system functionalities.

Frex
 
Hello,

The displayed dBV resolution is 0.5dBV, and it is also almost the accuracy of the measurement
in the -80/+10 range ( the AD8310 specs say +/-0.4dB). It work very well to many MHz.
Anyway, i don't have already made serious measurements on the instruments, i want to finish the software first.

Frex
 
Hello JC,

I must confess that i don't had much time these weeks to work on the project.
Anyway, the amplitude loop control start to work now, but some job need to be done before to get a good instruments.
I take also much time these to learn using ModelSim with the latest version of Quartus and writing test-bench files.
So, the learning curve is slow, but doing this for a real design is very motivating ! So no worry, work is in progress.

Frex
 
AT2380 DIY stepped attenuators news..

Hello,

After much work, all the CPLD software is now almost done.
Tracking mode of the attenuator work very well now.

The output voltage of the attenuator can be tracked between +10dBV to -85 dBV,
regardless of the input voltage or output load.
That's A very convenient way to fit with particular level for THD measurement.

I added also a new functionality that allow to display output dBV level
in both Manual and Auto mode.
The level measurement work within +/-0.5dB error beyond MHz range.

In both modes, the step size can be choose to 0.5,1,3,6 or 10 dB/crank.
A 1s push allow to switch from Manual mode to Auto mode.

I made few bandwidth measurements, and the frequency behavior is very good.
Even if performance verification is not easy.
I used the Analog-Discovery to send a DC-2MHz signal and analyze
output of the attenuator for -20,-40,-60,-80 and -100 dB settings.
The result is show below.

AT2380_Att-test-2MHz.jpg


We can note that higher is the attenuation level and noisy is the response.
This is a boundary of the Analog-Discovery ( 14bits ADC..).

Now, some results in a lower bandwidth range (DC-96kHz),
with my AK5394ADC in V2/V1 mode (ratio of Vout/Vin of attenuator).
The result below to different levels.

AT2380_spectrum.gif



We can show on these measurements that frequency response is very flat,
even much beyond audio frequencies.
I also made 10kHz single tone verification where it is possible to verify
much higher attenuation level.

To complete the design folder, i started to wrote the user manual of the project.
But as promised, i will publish a small movie to present the AT2380 soon.

Frex
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.