Re: Open HPSDR Hermes (14 bit ADC, EP3C25) Transceiver Card

Larry Dodd

Right on.  Not many amateur radio operators or radio enthusiasts are in a position to pay >$10k for a receiver or transceiver. In reality they don't "need" that level of precision even though its desired. Far surpasses the minimum requirements of the FCC rules. Yes its great to pursue and more power to those that are capable of doing so. Pure Signal is a great achievement.  Can't imagine the dollars spent on high end microphones and audio equipment that are far beyond what is necessary for voice communications. The ultimate is that 24k Gold Plated CW key.
Larry K4LED

K4LED Links:

On 2/25/2022 18:06:02, jdow <jdow@...> wrote:

There is an economics pressure here. Who is going to be the first company to give up being price competitive in a small commodity market and will that company survive the hit to its bottom line? A LOT of pressure from actual customers will be required to advance the commodity transceiver state of the art.

(And this thread is about the Hermes with it as a proof of concept for more general improvements - and their failures. It's been interesting.)


On 20220225 06:50:24, Conrad, PA5Y wrote:

Hello I never meant to come across as acrimonious, just insistent.


The plot is from a Rohde and Swartz FSUP signal source analyser in this case.


The LO I used is from Kuhne electronics and is known as the XO1, it was intended as a 116MHz LO for use with 144 MHz transverters. It is not a custom product. It is simple Xtal oscillator with a narrow band PLL. The plot shows its performance when locked to a Leo Bodnar GPSDO at 10MHz.


It is normalised to 1Hz so in an SSB BW you can add 68dB to this plot assuming 2400Hz BW for SSB. I do not agree that this performance levels are only obtainable by special equipment. In fact older equipment like the IC-202 used to fair better. New equipment is making a pigs ear of this and transmitters are getting worse. A K3S and a transverter is all you need, or indeed a FDM Duo, Hermes, Flex or ANAN SDR.


These radios are all capable of meeting the code of conduct which in practice does have a positive effect. Not perfect but its existence is useful.


Now that we have such excellent receivers maybe we should pay attention to the TX chain?


This thread was about the Hermes, I apologise for hijacking it.







From: <> On Behalf Of jdow via
Sent: 25 February 2022 14:30
Subject: Re: [SDR-Radio] Open HPSDR Hermes (14 bit ADC, EP3C25) Transceiver Card


Um, I am not sure I understand that picture. That is the analyzer's specification or something very custom you have put together? I'm answering both emails here.

If that is the noise level you are insisting everybody meet I suspect you are going to be a very frustrated gentleman for the rest of your life. (Some old rigs never die. They simply make more noise.) The IC7300 might be capable of being tamed with a reference oscillator replacement. But I bet more bits in the D/A are required for that design than exist in reasonably priced D/A converters. Regardless "state of the art" only appears in hand tweaked radios or VERY expensive military or scientific equipment where there is a perceived need. (And codes of conduct are not worth the bits they fill on an exabyte storage array. Cats and humans cannot be herded.)

Even if 7300s are not "clean" as you would have it are they cleaner than their peers on the commercial market? From your complaints I suspect they are not. But it is a point worth asking. I ignored the argument until it got acrimonious.


On 20220225 04:03:42, Conrad, PA5Y wrote:

Hi, no it is actually 5 different IC7300s tested by 3 different people, including Rob Sherwood. if you have time you can see the plots and some other data that I posted earlier just do a search. I’m in my lunch break or I would do it for you.


Yes my TX IS that good but it is easier for me because I use 0dBm transverter drives at 28Mhz and transverters with high quality Xtal local oscillators. The 0dBm from either my K3S or TS-890S is very good indeed. IMD3 from both is better than -50dBc.


On 144 MHz,  PN noise at 1kHz separation is -144dBc/Hz and close to 156dBc/Hz at 10kHz. This is with the LO PLL active, free running it is a little better. Composite noise is the same as in this case it is PN dominated. IM3 is -38dBc but more importantly the 7th and 9th orders are better than -80dBc. This is achieved by using tetrode finals on all bands.


432 and 1296 are of course worse as the LOs are multiplied from Xtals in the VHF region. However my TX is some 40dB better than an IC-9700 at 1.3GHz at 10kHz separation.



Only 50MHz is a little worse due to my TS-890S PA linearity but with the noise on 6m being higher it is acceptable. When I am on FT8 where linearity does not matter it is superb. But who cares we are all on the same ‘channel’.




Conrad PA5Y


From: <> On Behalf Of jdow via
Sent: 25 February 2022 11:49
Subject: Re: [SDR-Radio] Open HPSDR Hermes (14 bit ADC, EP3C25) Transceiver Card


I suspect your specification is still beyond the state of the art. Is your transmitter that good?

It sounds like the 7300 is cleaner than average close in but has some issues fairly far out. That is a little surprising. I wonder if the "IP+" technology is simply RF feedback to linearize the transmitter stages such as is seen in the old Collins kW SSB amplifiers. I can see that having an issue that depends on the charteristics of the feedback loop. But those are strange characteristics.

I take it the noise peaks are broadband, right? Do they depend on the transmit frequency in any way? It sounds very much like there is a defect in a specific radio near you. Is that the case or are hams measuring this in general? If this affects all the 7300s then the FPGA code will probably have to be reworked to fix the bug. I kinda wish I had one and a nice lab with lots of pertinent test equipment. I doubt Keysight and ICOM would set me up to do it. (It'd be better for them to set up a tiger team of their engineers with the equipment needed to suppress this possible bug.)


On 20220225 01:55:45, Conrad, PA5Y wrote:

By clean I mean - can I hear anything of sufficient magnitude to disturb the noise floor anywhere in band, and hence inhibit my ability to receive signals close to the noise floor. The noise floor in a semi-rural location on 28, 50 and 70MHz is not particularly low so I do not feel that this is unreasonable. The biggest problem with the IC7300 is the AM noise bump at 20kHz and 130kHz which is only 60dB down on the carrier when running at 30W on 28MHz. This power level is typical when driving an amplifier.  When modulated with an SSB signal this occupies a considerable amount of bandwidth.  The close in PN -s only -120dBc/Hz at 1kHz and  -130dBc/Hz at 10kHz. The composite noise is -115kHz at 10kHz, this is not what I would consider clean, even close in. The K3S( and a few others) is far better in this respect although the PA has quite poor linearity, especially on 6m.


The Hermes will do a much better job with any decent PA, even without pure signal. There are some spurs @ -65dBc which improve with a higher sample rate.  I have not checked these with the V2 firmware but I will do immediately I receive my Hermes from Apache.


I do not think that a transceiver at the IC-7300 price point has fixed pre-distortion. As you quite correctly say keeping it under control over temperature and voltage variation would be difficult. Also Icom would almost certainly have mentioned it in their advertising. What the IC7300 does well is manage audio overshoots by using a ‘look ahead’ algorithm. This is maybe where the impression of it having a clean TX comes from.


I think that the problem here is that so many transceivers are quite bad, so I can just about accept that the IC-7300 is less bad.




Conrad PA5Y


From: <> On Behalf Of jdow via
Sent: 25 February 2022 10:27
Subject: Re: [SDR-Radio] Open HPSDR Hermes (14 bit ADC, EP3C25) Transceiver Card


OK, please refresh my mind what you mean by "clean". It certainly appears to be sending something much closer to pure signal concentrated within its intended bandwidth than most other transmitters. I also understand that it is not as good as it can be with full active predistortion. Both might generate increased noise at some significant separation from the intended signal frequency. I am not sure of the mechanism by which this would take place.

As an ornery critter there is nothing I would sit down and declare "clean" without a definition of "clean". I guess I am asking for your definition of the term or even of the term "clean enough".


On 20220225 01:13:11, Conrad, PA5Y wrote:

I KNOW for a fact that the IC7300 is not clean. Please provide me some evidence to the contrary. I have measurements with a R&S FSWP Signal analyser out to a 1MHz BW. You guys are looking in an SSB BW which is fine unless you happen to be 130kHz away, then you will hear plenty of splatter from the IC7300. I tested the IC-7300 as a result of hearing this on 50MHz. In other words the lab tests were driven by on air experience.


You are spreading misleading and incorrect information.




Conrad PA5Y


From: <> On Behalf Of oldjackbob via
Sent: 24 February 2022 23:12
Subject: Re: [SDR-Radio] Open HPSDR Hermes (14 bit ADC, EP3C25) Transceiver Card


On Wed, Feb 23, 2022 at 12:24 PM, Max wrote:

I just think if we follow good design and signal chain management that superb, clean signals can also be generated without the need for PS, that's all.


I disagree. My signal is my signature. Audio clarity is admittedly important but what matters most to me is a splatter-free signal, and in that regard nothing comes close to what PS can produce. As I stated earlier, even the best Class A amplifiers only have -40dB splatter. PS starts at -60db and often pushes -70dB. That means that PS reduces splatter to less than 1% (and at times only 1/10 of 1%) of what even the best traditional equipment can produce.

I also agree with Simon that the 7300 (and also the 7610 for that matter) produces an extremely sharp-edged signal with unarguably less splatter than is achievable with any other non-ANAN radio. It was explained to me by Ray N5LAX that the reason those radios are so clean splatter-wise is because ICOM embedded an algorithm into the firmware that functions very similarly to PS in the sense that it corrects for non-linearity in the PA, but it is a fixed-value (i.e., static) correction only, not a real-time self-adjusting correction value such as is employed by PS. NOTE: I have not been able to verify that info, but that is what was told to me over-the-air by Ray when I commented that I had no idea how the 7300 could produce such a clean, sharp-edged signal. So I'm repeating it here.

IMO it's unwise and short-sighted to summarily dismiss the game-changing value that PS brings to the hobby. Its benefits are so unarguable that I (and Rob Sherwood) often wonder why the big-name radios don't offer it as a standard feature. It's free, so utilize it for goodness sake!






Join to automatically receive all group messages.