AD9866 Alternatives

[Steve Haynal]

Hello List,

This week I was looking at what alternatives to the AD9866 exist, where an alternative is some combined ADC/DAC suitable for a HF transceiver that costs less than $50 per single piece. Here are some possibilities:

http://www.mouser.com/new/Analog-Devices/adi-ad9963-mxfe-converter/

http://www.mouser.com/new/Texas-Instruments/ti-afe7222/

http://www.ti.com/product/afe7225

All of these are a bit more expensive than the AD9866, but include dual ADC and DAC. Dual ADCs are attractive for true IQ baseband processing, diversity, and increasing dynamic range through averaging. I'm not seriously considering replacing the AD9866, but am curious to see what else is out there. What do other people think?

I am surprised that no commercial company with more engineering time and resources hasn't built a SDR around the AD9866 or one of these other chips yet. Perhaps one is on the way.

73,

Steve

KF7O

[ok1...@gmail.com]

Hello Steve,
I guess the reason for change would be the opportunity to reach 50 MHz or 70 MHz bands...
73 es GL
Tomas, OK1DXD, OL2W

[James Ahlstrom]

These are interesting chips, but seem basically similar to the AD9866; 12-bit Tx/Rx.  They are designed as radios instead of a power line modem, so maybe the performance is a bit better.  More compelling would be a chip with 14 or 16 bits.

It is hard to judge performance without using them.  And it is hard to chase new chips and still get a product out the door.

Jim

N2ADR

[Alan Hopper]

Steve,

all very interesting, I think it is essential to keep an eye on the alternatives as this project is much more than just the ad9866 and it would be a shame to be caught out by better options becoming available or the current one going out of production.  The over sampling performance of these look interesting with high quality 2m looking possible. Twin adcs is tempting but I'd go for the cheaper single adc and get two if that was my interest ( just so happens that a friend of mine G3NPC has got me excited about using 4 sync'd receivers for a 4 square antenna).   I believe your softrock price point goal is still a good guiding light rather than being tempted to up the spec and price.

I've been thinking about the user interface for sync'd diversity receivers, you could have a number of virtual receiver sets each focused in a different direction and then play them from the right direction through a surround sound system.  Connected to a microsoft hololens you could then turn you head to hear the signal of interest and be presented with your log for that contact.

73 Alan M6NNB

[Steve Haynal]

Hi Jim, Alan and Tomas,

Thanks for the comments. I agree 14 or 16 bits would be more compelling. Several months ago I asked Analog Devices to release an update to the AD9866 with 14 bits. Since we understand many of the AD9866 quirks, it makes sense to stick with it.

73,

Steve

KF7O

[in3otd]

The TI devices have also an internal NCO and a simple /2 decimation filter (besides other auxiliary stuff), might help in reducing the resource usage of the FPGA. The AFE 7222 is not much more expensive than the AD9866; worth a closer look after finishing the first H-L version with the AD9866 :)

73 de Claudio, DK1CG

[Dave Clarke]

Steve:

Looking at the Hermes Lite 2 and the current use of the AD9866 and its realized performance is excellent, however, a transition to a higher bit rate chip with all the capabilities of the AD9866 will  require higher capacity FPGA.

A redesign of the PC board to accommodate the new chips BUT I suggest even with the existing board, There is a need to ensure no components can sheared off Like I did with C56 by a assembly error (Clumsy Ham Proof) 

The Hermes Lite 2 is a fantastic CW Portable Radio when paired with a Raspberry Pi 5 and a touch screen.

My personal observations can see a higher bit rate device may allow further development into weak signal detection while living in High RF Noise environments (my problem).

the higher bit rate may enable a higher dynamic range which is essential for potential operation adjacent to Urban High RF noise levels where the majority of us live. Few can afford a place in the quiet rural areas, or our work requires us to live in the Urban environment.

All This Aside, The Existing Hermes Lite 2 with appropriate Software does an absolutely amazing job, and fills a an affordable and valuable radio resource.

Thank You for your efforts. Dave VE6LX

[Clifford Heath]

Higher bit *depth* gets you more dynamic range. Increasing the sample rate increases the maximum frequency you can work.

The cheapest alternative to the AD9866 is two or three chips which each cost as much as it does. And takes more PCB area of course.

I'm not saying we shouldn't go there. I'm saying that you're proposing an entirely new design, and a lot of work. You wind up reconsidering every single design decision, and reviewing the entire space of adjacent competitor products.

The end result isn't an HL2 in any real sense. It's just a new SDR.

Clifford Heath 

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/hermes-lite/3fea22a5-8939-4a72-824d-b6d84dc2acbbn%40googlegroups.com.

[Steve Haynal]

Hi Dave,

Yes, there are definitely better ADC/DAC options out there compared to the AD9866. But I have never found one as cost-effective as the AD9866, and I have spent considerable time looking. My goal was never to make a high performance SDR. The openhpsdr group already did a very good job of that. My goal was to create a low cost yet decent DDC/DUC SDR. People like the HL2 and then start discussing how to make it better, but never fully understand that its strength is the simplicity which keeps its cost down.

73,

Steve

kf7o

[Ed Marciniak]

Assuming oversampling is done correctly, you get additional bit depth when reducing the output sample rate.

In order to allow an FPGA configuration to remain static, you really need more than 16 bit wide paths, along with >16 bit depth after the first filter/decimate or two. If you want the FPGA design to permit swapping in a 16 bit ADC to start with, the the entire pipeline will have to be wider and than 16 bits.

There’s still ADC noise floor, alias filters and some other things to consider.

It’s useful to think in terms of measured power spectral density noise floor normalized to 1Hz bandwidth, the point at which the ADC begins to saturate and how much spectrum you can work with at a given noise floor.

---

From: herme...@googlegroups.com <herme...@googlegroups.com> on behalf of Clifford Heath <cliffor...@gmail.com>
Sent: Sunday, February 15, 2026 3:42:26 PM
To: Dave Clarke <ve6...@gmail.com>
Cc: Hermes-Lite <herme...@googlegroups.com>
Subject: Re: AD9866 Alternatives

 

To view this discussion visit https://groups.google.com/d/msgid/hermes-lite/CAPnXPY2szHBhLJG0FtDJqzszSURW5BfGKGvAO8UMDcGxF65QdA%40mail.gmail.com.

[Clifford Heath]

On Mon, Feb 16, 2026 at 5:29 PM Ed Marciniak <edr1...@gmail.com> wrote:
> Assuming oversampling is done correctly, you get additional bit depth when reducing the output sample rate.

I had hoped no-one was pedantic enough to make that argument.

To increase sampling rate from 76.8MHz enough to get *even one more
bit*, you need to go to 150MHz, or to 300MHz for two extra bits. And
even then it only works if the dithering noise spectrum and amplitude
is just right.

Have you looked at the price/performance ADCs in range this sample
rate range, compared to just using a 14-bit converter at the current
sample rate?

It's a theoretical argument that is much more rarely useful than folk
seem to think it might be.

In any case HL2 operates in the HF range, with front-end filters,
where atmospheric noise dominates. It's all very well for KiwiSDR etc
to use 16 bits, because they usually operate with no filters,
capturing the entire HF band in one go. Extra dynamic range is quite
useful in that case.

Clifford Heath.

[Ed Marciniak]

I would find it extremely useful to have a transceiver that could run in the first Nyquist for 2m without a transverter and either populate a filter with alternate components or a include a vacant optional set of pads for another band(or bypass). I’d also bypass or not populate the 10W PA as I want no more than +10dbm out.

I’d be using it for portable microwave operation. Phase noise is critical. The ability to lock directly to 10MHz is important. It must handle nearby interferers at -10dbm tens of KHz away.

I’ve run into situations where another station was transmitting 8W 10m away (10GHz) while I was in RX  listening for distant stations perhaps 10db over the thermal noise floor. With some conversion gain from a 10GHz to 2m and a 2m to 10m tranverter.

---

From: Clifford Heath <cliffor...@gmail.com>
Sent: Monday, February 16, 2026 1:58:38 AM
To: Ed Marciniak <edr1...@gmail.com>
Cc: Dave Clarke <ve6...@gmail.com>; Hermes-Lite <herme...@googlegroups.com>
Subject: Re: AD9866 Alternatives

 

[Ed Marciniak]

The problem with the openhpsdr group is the lack of any currently available hardware.

---

From: herme...@googlegroups.com <herme...@googlegroups.com> on behalf of Steve Haynal <softerh...@gmail.com>
Sent: Sunday, February 15, 2026 7:46:11 PM
To: Hermes-Lite <herme...@googlegroups.com>
Subject: Re: AD9866 Alternatives

 

--

You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/hermes-lite/e35a11ea-56cc-4c13-870b-10f50d842824n%40googlegroups.com.