attenuator - response summary

[Paweł Kuśmierek <pawel.kusmierek@xxxxxxxxx>]

Hello everyone,

 

thanks to everyone who replied to my question about attenuators.

 

Some responses were off-list, and some were "off-off-list" (I asked elsewhere), so here is the summary.

 

To remind you, I was looking for a "commercially available precise digitally controlled (i.e., repeatable) attenuator with fast and quiet attenuation changing. Preferably, but not necessarily controlled via RS232."

The RS232 was suggested because I want it to be controlled from CED Spike2/Power1401 system or from NBS Presentation, or from similar systems, and RS232 looked like a sensible (although pretty old) solution.

 

Basically there were two suggestions.

 

1. TDT PA2

 

advantages:

- wide attenuation range of ~120 dB (0 to - 120dB),

- small gain step (0.1 dB),

- high accuracy (0.05 dB),

- good noise/distortion figures

- manual control possible

- attenuation value is displayed

- ready to use

 

disadvantages: 

- price: $725 for a single channel device, but additional control/power/PCI cards, etc. hardware and software are needed, which drives the price of a single-channel attenuator to ~$2,500. I think that the second channel can be added for additional $725 only, but still it is not cheap

- control: via an AciveX control, I am not exactly sure if (and how) I could do this from CED or NBS systems.

 

A person suggested TDT RX5 and I was not able to find what it is. However, RX6 and RX7 appear to be powerful DSP devices, so RX5 is probably too much capability-wise and price-wise than I need. Please correct me if I am wrong here.

 
2. A "digital potentiometer" chip. Specific models suggested were Dallas Semiconductors DS1801, Cirrus CS3310 (or 3308 for 8 channels) and Texas Instruments PGA2320.

advantages:

- low price - the chip itself is in $5-$20 range, for a 2-channel device

- wide attenuation range ~127 dB for the CS and PGA (~63 dB for the DS)

- pretty small gain step/accuracy (CS: 0.5/0.05 dB, PGA 0.5/0.1 dB, DS 1/not sure)

- good noise/distortion figures

- control via three serial TTL lines (more may be needed if certain features cannot be set to enabled or disabled permanently) - this could be achieved from NBS Presentation via LPT printer port or NI I/O board, and from CED Spike2 via the 1401 interface

- zero-crossing mode - the chips may be set to attempt to change atttenuation on analog zero-crossing to reduce artifacts

 

disadvantages:

- a device has to be built around the chip - power supply (PGA: +5V, +15V, -15V, CS: +/- 5V, DS: +3 or + 5V), connectors, possibly buffers. This will increase the total price, but definitely it will be still well below TDT PA5 price, even with labor factored in.
- for CS and PGA the 127-dB attenuation range is NOT 0 to -127dB, it is +31.5dB to - 95.5dB, which may have clipping implications 

- no manual control and no way to read back the set value (unless special steps are taken)

 

One person warned me that "digtal audio attenuators" typically do not have 1 dB steps, but fine steps at high levels and coarser steps at lowe levels. However, the chips that were suggested to me have 0.5 or 1 dB steps over their attenuation range.

He also proposed using and old SoundBlaster SB16 or AWE32 as an attenuator. These cards need a computer with ISA bus, which are not available for me (and frankly, I doubt there are many operational machines from these days around).

 

 

Thanks again to everyone for your suggestions,

 

Pawel