Radioberry Preamp supply weakness

[Samuel Lourenço]

Hi,

I found out that this board really can't be fed with more than 12 V. This is a big no-no, because I need it to be fed with 13.8 V for portable operation. However, no matter how I try it, the ST1S10 blows, sometimes immediately, sometimes right after going to TX. I found no problems when feeding 12 V into it, though. Mind that I've got these from Mouser.

I had hiccups with devices from ST Microelectronics in the past. So, what I would suggest is either replace the DC-DC converter with a more robust one, say, a TI TPS54233. Although not exactly pin compatible, it is impossible to go wrong with it. I tested it inside a tracker supply, which is a very critical application and messes up the feedback loop of the converter quite a bit, and it held without any craziness.

Another option is to just get rid of the DC-DC altogether, and just get a pair of more robust transistors. Say, use a AFT05MP075GNR1, which might be overkill but that's precisely what you want. And also get rid of the temperature sensor and whatnot while at it, so you can have more space for an heatsink, like the CUI HSB25-282810.

Anyway, take my suggestions at face value, but I think the Preamp board could only win by being more simplified.

Kind regards, Samuel Lourenço

[Dan Habecker]

Samual,

Thanks for this info.  It turns out I experienced the same problem with testing my preamp board.  The ST1S10 let the magic smoke out.  After seeing your comments I decided to switch to theTPS54233.  I created a small board using KiCAD and will just patch it in.

TNX & 73,

Dan W9EQ

[Samuel Lourenço]

Hi Dan,

Could you then provide that board or the design in some form?

Kind regards, Samuel Lourenço

> --
> You received this message because you are subscribed to the Google Groups "Radioberry" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to radioberry+...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/radioberry/a3b96f49-40af-4ad1-880f-308bfd7329cfn%40googlegroups.com.

[Marc Jatteau]

Hello Dan, 

I'm also interested in the Kicad file. Can you send it to me? Thanks.
73 by Marc F6GKG

[Dan Habecker]

Hi,

Here are the files.  I haven't recieved the board back yet, so it is untested at this time.  I'll give some results when I get them back and tested.  The file name has a sort generic name since this was my first go at using KiCAD to lay something out.  Hope they are of some help.

73,

Dan W9EQ

[Samuel Lourenço]

Thanks, Dan!

> To view this discussion visit https://groups.google.com/d/msgid/radioberry/6c56f71f-ad75-44f5-8020-08c931d2ae42n%40googlegroups.com.

[Dan Habecker]

I got my PCBs back and built one up, and put in into the circuit.  So far everything is looking good.  I have the input voltage set to 13.8V and I have ran the preampilifier to drive a MRF101 amp to full power.  No magic smoke and it holds the 8V regulated to the preamp just fine.  Oh by the way, as far as the inductor goes, I just took the off the PCB (it had to come off anyway) and reused it.)  I used a footprint that would allow me to reuse it and the value was valid for the new DC-DC converter.

[Samuel Lourenço]

Great news Dan. I've kept my Radioberry Amp PCB just in case, although I had removed the connectors. So I guess I will have to install them again.

One thing though, one of the pads got lifted when the previous DC-DC converter released its magic smoke. So, is that module of yours solderable on top of the PCB without needing any pads that were used by the former DC-DC?

Anyway, I'm happy that one of my suggestions is valid. I wouldn't expect otherwise, because the TPS54233 is a solid choice.

Kind regards, Samuel Lourenço

To view this discussion visit https://groups.google.com/d/msgid/radioberry/1ac812ed-0ad9-4049-903b-f74251afcbfcn%40googlegroups.com.

[Mario Vano]

Me Too!

I've gone through 2 regulators before I figured out the design was faulty - I may try an inductor that doesn't saturate at a lower current than the power amp requires at 80 meters or a pre-regulator.

(Or I could just be more careful)

AE0GL

M

[Samuel Lourenço]

I sincerely don't think that the inductor alone is at fault. Any DC-DC
regulator worthy of that name should not blow and take the PCB with it
(as this last one took, because the current was not limited) case the
inductor saturates. Conversely, the inductor is indeed too small. I've
designed a 2 amp supply with a TPS54233, and the inductor that I've
chose for that is 8 times bigger in volume, and it still saturates if
the output gets shorted. You can hear it whining. However, the DC-DC
never blew or anything. In fact, I've only had one DC-DC converter
from TI die on me, and it died silently after much abuse (over-current
and over-temperature for long periods).

Kind regards, Samuel Lourenço

> To view this discussion visit https://groups.google.com/d/msgid/radioberry/58a46ef0-c2ab-4766-9919-6519c9749986n%40googlegroups.com.

[Mario Vano]

all absolutely true, Samuel! I just can't help poking at it while it's in front of me! It's become an obsession.

I suggest the QRPLabs 10 Watt Linear as a cheaper and better replacement (after you throw away the ugly heatsink and replace it with something smaller). I've got several and they're much more robust and deliver full power when driven by RadioBerry!

M

[Samuel Lourenço]

Hi Mario,

I actually have bought a 10 W amplifier from 60dBm.com (this one: https://www.60dbm.com/product/10w-broadband-hf-6m-1-55mhz-rf-power-amplifier-cased/). These are robust as well, and also have a TX on input, which comes handy for a board I'm doing for the Radioberry, which is basically a TX/RX board. My board will also allow other things, such as giving a boost to the 5V power to the Raspberry Pi and also supply 5 V to a touch screen.

Regarding the Radioberry amplifier, it doesn't deliver what it is expected and it is not well rated for the job.

Kind regards, Samuel Lourenço

To view this discussion visit https://groups.google.com/d/msgid/radioberry/f5b30997-e73f-40b6-93ff-fb4d7b52e1b3n%40googlegroups.com.

[Mario Vano]

So, I've been reviewing this in detail and I'm frankly puzzled!

Almost all of the Radioberry preamp board schematic and BOM appear to be identical to the latest Hermes Lite 2 schematic and BOM. In particular, I see no significant differences between the HL2 use of the ST1S10PHR regulator and all of it's surrounding components - including the inductor and those of the preamp board. I know what parts are used on the Preamp board because I built it myself and purchased all the parts from Digikey using the matching part numbers. There are no "cheap" or cloned substitutes.

The HL2 is specified at allowing 12-16 Volts input, however! While I'm not willing to experiment extensively with my own HL2, I'm sure I've run it from 13.6 Volts in the past with no problems at all. Furthermore, the HL2 uses more than one ST1S10PHR and all seems to be well with that.

Any Ideas? Perhaps it's a layout issue? Perhaps I've missed something about the HL2 published schematics and BOM (from the Github).

Any constructive ideas as to possible causes for the discrepancy would be welcomed. Yes, we all agree that other power amps and regulators are better, but I'm very curious what's going on here. It makes no sense!

thanks in advance,

AE0GL

Mario

[Samuel Lourenço]

Hi Mario,

I can confirm that all parts that I've used were genuine, and sourced
from Mouser. I've managed to blow three ST1S10 DC-DC converters.
Always with 13.8 V and always not immediately after plugging in, but
only when starting TX. However, I'm aware that quite a few people had
the same issue of the ST1S10 getting damaged, as reported on other
threads. This is an issue that should be looked at.

Kind regards, Samuel Lourenço

> To view this discussion visit https://groups.google.com/d/msgid/radioberry/00240060-3206-41b6-b8c3-68f15cbfa0d3n%40googlegroups.com.

[Mario Vano]

Hi again Samuel:

I think we all agree on the symptoms, but it may be tricky figuring out what's causing it, since blowing regulators is not a good troubleshooting method! I'm on my third one as well - as I said, I don't think the PCB can handle many more hot air removal and resoldering cycles!

I think we need to focus on the differences between the Hermes HL2 and the Radioberry and try to figure out and measure whether those could have anyting to do with the problem. It's going to take a while, and I only have so much time to devote to it, but I'm going to be working on it over the next few weeks.

Here are some of my current thoughts. If you're equipped to test some of them it would help all of us:

1. I've had trouble with the FPGA outputs being in a half-on, half-off state at power up until the array is loaded. The Hermes does not have this problem! It's gate array is ready much sooner. Radioberry leaves the line that control the bias and power amp switcher in a floating state for a very long time during bootup and the situation is not resolved until the gateware and driver are loaded by the operating system.. I don't know what this could harm, but it needs to be explored.

I suggest we need  to try to capture the details and relative timing of these signals (and perhaps others) during power up and see if anything nasty could happen.

Interestingly both the HL2 and the RadioBerry power amp have pull down resistors on SOME of the signals - but on others they're installed but labelled DNI which means they are not present.

2. Similar problems could be occurring during PTT and the same behavior needs to be captured and analyzed for any possible timing glitches.

3. The bias circuitry seems to be broken and cannot be written. Writes to store the bias values silently fail - unless you manually load the driver and watch for messages when Quisk and other programs run. Sadly, the driver reports data about what is written when it SUCCEEDS, but not when it FAILS - which is what we need to know. It will take patching the driver to find out what's going on.

In any case, the bias is always completely wrong and possibly undefined except briefly during the calibration process in quisk and sparkSDR. On my board, the rest of the time one pot is defaults nearly to 0 and the other appears to be set to a different value than when Quisk has just set the bias!

The power amplifier is capable of being biased to draw more current than the switching regulator is designed to handle. If it is ever this state before the regulator is turned on, the regulator might be driving a load that's destructive to it - the data sheet has several fixes for other ways this can occur, but none of them would help this scenario.

Unfortunately, it's very hard to monitor this behavior without modifying the programs to provide better trace information. That's doable, but all this takes time.

4. There could be problems related to layout of the regulator. That's REALLY hard to  troubleshoot, as scoping various points usually disturbs things enough to provide false information.

All these are hypothetical and none of them may be occurring or causing the problem - but it this point I can't see any way to proceed except to eliminate them systematically.

This may take a while...

Furthermore, I'm not convinced this board deserves high priority! I'm pretty happing running 3 Radioberry based radios here using other power amp designs that are more straightforward! I'm still curious about the problem because i've invested a lot of time and effort into one of these boards because I needed it to test compatibility with my own LPF filter board before I publish the design. So far the LPF board works flawlessly, with the preamp, but the preamp's problems have been a major distraction!

To summarize:

I'm pretty sure component choices are NOT the problem as the schematic and BOM is IDENTICAL the HL2 and it does NOT have this problem. The inductor is exactly the same part as in the HL2.

We need to capture the startup of the board and analyze where that could go wrong.

We need to do the same with the T/R switchover sequences for all modes that might cause problems.

We need to find out what is wrong with the bias circuit as this is CLEARLY not working as intended and this might have side effects.

At some point it may be constructive to compare all the above behavior on an HL2 - but I'm reluctant to do that as I'm having too much fun using it and would hate to tear it apart to start using it as a reference board - it would mean leaving it unusable until I find the problem with the preamp board.

I'll keep working on it, but it may take a while. I came back here to this old thread to see if anyone had any other ideas of where to look, and I'd still love to hear any!

In the meantime, the best thing is to run the board from a well regulated 12V supply. Fortunately it's easy to get USBC PD battery packs and wall warts to do this reliably.

Hope you or someone else reading this can help by analyzing the problem or even running some tests of their own.

73 for now,

AE0GL, Mario

[Mario Vano]

Two more things:

5. On power upof my board, the 12 Volt supply consistently shows a current drain of 14-15ma here. This persists until the gateware and driver are loaded and then it drops to 0. It's probably still got some residual current for the I2c devices, but that doesn't show on my power supply's meter. I believe that this means that it is NOT starting up as it should..

6. If the board gets into an indeterminate state - either at startup or at any other time during operation, there is some chance that the power amp can be enabled but that the antenna relay is not enabled leaving it with no load. I believe that this can cause the amplifier to oscillate if it happens and the current drain could be very high. Hot switching the bias supply, or antenna relay while this is happening could also be causing unpredictable transients....

M

[Samuel Lourenço]

Hi Mario,

I think I was misunderstood. Of course blowing DC-DC is not a good way
to troubleshoot. Neither we need to, because it is evident for me that
the conditions to blow up a ST1S10 are consistent:
- 13.8V supply;
- Start TX.

However, I disagree with you when you mention that the ST1S10 is not
the culprit, because that doesn't happen in HL2. The fact that it
"reportedly" didn't happen with the HL2 (as far as we know) doesn't
mean that the design is not fragile. A DC-CD converter could present
some issues like increased noise due to bad layout, but bad layout
alone doesn't explain why it blows up consistently under certain
conditions. Clearly the design is poor, but so is part selection.

I wouldn't be surprised if the lack of a proper freewheeling diode at
the relay coil terminals would create a cascading failure, because
even components that are on the 3.3V rail go when the DC-DC converter
blows up. And I'm aware that the transistor that activates that relay
has protection, but it won't protect any other components except
itself.

Anyway, clearly replacing the DC-DC converter for a more robust one
did the trick, but this design is clearly subpar and has so many
problems that, for me, is not worthy of any redemption. I'll be using
another solution: a relay board and an external amplifier.

Kind regards, Samuel Lourenço

> To view this discussion visit https://groups.google.com/d/msgid/radioberry/839c3133-11da-442f-b167-5a9973eb71f2n%40googlegroups.com.

[Mario Vano]

Samuel, the relay driver part was designed for this purpose and has built in diodes/zener diodes and I'm sure they work well (if the layout is proper). I suspected the same thing myself and spent some time investigating it. I don't think that's the culprit. I've spent a lot of time on this already and I only am mentioning the leads that I haven't yet followed up!

By the way, I forgot to mention that one more thing that needs to be checked is to confirm that the actual data being sent to the bias control pots for programming them is correct by spending some quality time with the I2C analyzer on the scope...

I agree with everything else you said (as I have from the beginning), and use other PA and LPF modules myself. I think the board is very hard to heatsink and at best cannot deliver reasonable power for standalone use. Unfortunately it appears that there may be at least 50 of these boards in the field right now (not counting mine) and presumably some people would like to use them so it would be nice to find a solution.

It may turn out to be as simple as a small change to the driver or FPGA code or adding a resistor/capacitor in the right place.

As I said, I still intend to pursue this, but not at high priority at the moment. Winter is coming, however, and that's usually a good time for projects like this...

M

[Samuel Lourenço]

As far as I'm concerned. I will not spend any more time on this. It is
really not worth pursuing that avenue. I will not spend another penny
on a flawed board that is over-engineered, subpar, hard to assemble
(due to the 0603 components) and does not deliver.

For me, the Radioberry Preamp is pretty much gone. I will just
desolder a few more components and toss the board. It is a waste of
time and money, and I'm not happy about it. There are much better
venues that I have the intention of pursuing and they do deliver.

Kind regards, Samuel Lourenço

> To view this discussion visit https://groups.google.com/d/msgid/radioberry/12d60833-47bf-4311-bb67-db836f681702n%40googlegroups.com.

[SNAIL]

I accidentally found this kit in the recommended ones on YouTube. Big brother taking care of us. 

 Looks similar to the 60dBm amplifier but you avoid risky international shipping, if you are in USA of course.

20W Amplifier Kit – T41-EP SDT – AI6YM

[Mario Vano]

Hi Snail!

That's a really nice clean looking kit! I think it's part of the T41 project - which is noted for it's careful engineering. I couldn't find the dimensions anywhere - did you happen to buy one? If so could you report back?

I Generally use the Excelleyt QRPLabs 10Watt Linear amplifer kit. I have 5 of them in my various radio builds and they work great, are very clean, are very robust and actually very tiny if you discard the heat sink that comes with them and mount them on a smaller plate. Two of those radios are RadioBerry based - the rest are based on my older "RadioHat" board. The 10W Linear amps from QRPLabs are also quite inexpensive - or they were the last time I bought any!

I don't know how much it costs to bring them in from Turkey now...

Thanks for posting this...

M

[SNAIL]

My amplifier is on the way and will arrive probably next week. Will test it and post review with dimensions and better pictures when it arrived. I plan to use it as 10 W amplifier mounted on the back of transceiver box. Heat sink outside the box amplifier board inside.

[Samuel Lourenço]

This is what I will be using with the Radioberry to control TX/RX. RX
goes to the Radioberry and TX comes from the amplifier. There is an
RCA jack to enable the amplifier as well.

Kind regards, Samuel Lourenço

> To view this discussion visit https://groups.google.com/d/msgid/radioberry/cf46f97e-cb9e-423f-9b2e-b3586c2fd18bn%40googlegroups.com.

[Mario Vano]

Nice!

Perhaps we need to start a gallery of complete radio builds with block diagrams...

Everyone's doing it differently and that's a good thing!

Other's might be interested in our varying ideas - I have 3 complete radios now, and no two are built from the same boards!

M

[Samuel Lourenço]

Hi Mario,

Attached is the preliminary schematic. I've based this on the "hat"
for the Radioberry, since if the amp is not detected via SPI, and thus
only one line works for TX/RX switching. So, I use the same one line
to activate the relay and the external amplifier to be connected via
the RCA jack. The other lines from the Radioberry are used for supply
and to keep compatible with similar boards (again, based on the "hat"
board). However, it seems that the one line I use for TX/RX switching
is the one that always works as MOX regardless of what was detected.

This board will also provide a means to reinforce the 5V from and to
the RPi, via its USB connector. If you are using this with the RPi 5
or any version that supports PD, it is advisable to remove F1 (you can
remove the connector and the cap nearby for good measure). There is
also a header that supplies 5V to an external display (such as the
Touch Display 2). Surely you will need the extra USB connection is you
are using the display, plus the Radioberry, plus a USB sound card that
is much needed. You can run all without the extra cable, but "low
voltage" warnings will appear.

Kind regards, Samuel Lourenço

> To view this discussion visit https://groups.google.com/d/msgid/radioberry/d0df33ac-5c6c-49bf-8979-9b6c31a922c0n%40googlegroups.com.

[SNAIL]

My kit is arrived. I'm very happy with it. Nice circuit board with all SMD in place. Need to wind transformers and install transistors. All hardware included. Dimensions are 150x60x25 for heatsink. Board is 140x50.

You received this message because you are subscribed to a topic in the Google Groups "Radioberry" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/radioberry/P5oWWlHju50/unsubscribe.
To unsubscribe from this group and all its topics, send an email to radioberry+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/radioberry/CAEJA4F3Hp-SkhaRVSmeCn3c%2B-gqQW%2BJ5Nd66fTLtUu1Kv_baMA%40mail.gmail.com.