Closed jbqubit closed 6 years ago
@gkasprow When will the prototypes for this arrive?
@gkasprow List of suggested tests posted above. Let me know if that sounds okay, and if you can think of anything else/different we should do...
@thomasRF Please use English. For now, using google translate...
- Check temperature regulation, and thermal gain compensation. Remarks: we have a climate chamber? Do we take a refrigerator with a heater? ;)
Testing at ambient lab conditions should suffice.
Some additional test suggestions:
@thomasRF @gkasprow Note that these are not intended to be production tests that are performed on each unit that is shipped, but rather methods of verifying the design of the prototypes.
Sorry, it was internal communication and Thomas replied to all :)
Status update: mechanics is ready for integration
Looks good @gkasprow! Can't wait to see it in action.
Thanks to @gkasprow and his team!
@hartytp does the amps amping? I can has gains plz?
@hartytp does the amps amping? I can has gains plz?
I believe so. This particular Christmas Present is still in the post. @gkasprow has the data I believe.
Yes, the channel isolation is better than planned. Gain flatness is also better. Tomasz is processing all measurement and promised to send them all to @hartytp
more photos:
Ace!
Apart from the electrical tests above, we should check mechanical things like:
Anything else?
One final thing, it would be nice to have an actual name for this board rather than just "RF PA". Thoughts? No Russian names, please? Looking for a bombastic name implying "loud" and fitting with the "er" naming scheme...
First attempt:
Etna?
From: hartytp [mailto:notifications@github.com] Sent: Thursday, December 21, 2017 4:02 PM To: m-labs/sinara sinara@noreply.github.com Cc: thomasRF tomasz.filipek@gmail.com; Mention mention@noreply.github.com Subject: Re: [m-labs/sinara] RF PA v1 errata (#251)
One final thing, it would be nice to have an actual name for this board rather than just "RF PA". Thoughts? No Russian names, please? Looking for a bombastic name implying "loud" and fitting with the "er" naming scheme...
First attempt:
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/m-labs/sinara/issues/251#issuecomment-353372542 , or mute the thread https://github.com/notifications/unsubscribe-auth/AVx8azpenrqw8217NVT9mv3yUZBq113Rks5tCnL0gaJpZM4O0IaF .
As in the active volcano? Points for bombasticness. Maybe a little overly destructive for lab equipment?
@dtcallcock something from the audio world?
I thought about that as a source of power or vitality. By the way, probably the volcano is the place where Hephaestus had one of his forges.
From: hartytp [mailto:notifications@github.com] Sent: Thursday, December 21, 2017 4:32 PM To: m-labs/sinara sinara@noreply.github.com Cc: thomasRF tomasz.filipek@gmail.com; Mention mention@noreply.github.com Subject: Re: [m-labs/sinara] RF PA v1 errata (#251)
As in the active volcano? Points for bombasticness. Maybe a little overly destructive for lab equipment?
@dtcallcock https://github.com/dtcallcock something from the audio world?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/m-labs/sinara/issues/251#issuecomment-353380204 , or mute the thread https://github.com/notifications/unsubscribe-auth/AVx8a6aFI_l_36I3ZtS8rSyY-IV2JFYJks5tCnn_gaJpZM4O0IaF .
I prefer Etna to Gainer. Some other "-er" suggestions
Some non-"er" suggestions:
a few ideas for acronyms (http://acronymcreator.net/):
MoCHA Multi CHannel Amplifier MUsCA MUlti Channel Amplifier HaIFA HIgh Frequency Amplifier MoCAP Multi Channel AmPlifier MODAL MODular AmpLifier MaCAo Multi Channel Amplifier MUsCAt MUlti Channel Amplifier ReFRAME Radio FRequency AMplifiEr MARiA ModulAR Amplifier MICAM MultI Channel AMplifier MICRON MultI Channel RadiO frequeNcy MaCRON Multi Channel RadiO frequeNcy MARADONA ModulAr RADiO frequeNcy Amplifier MUsICALE MUltI Channel AmpLifiEr REPERA Radio frEquency PowER Amplifier MARQUA ModulAr Radio freQUency Amplifier REPEAL Radio frEquency PowEr AmpLifier MARINA ModulAr RadIo frequeNcy Amplifier REPAIrER Radio frEquency Power AmplifIER MURIRAI ModUlar RadIo fRequency AmplIfier
Definitely Maradona, eh English folks? ;)
I thought about that as a source of power or vitality. By the way, probably the volcano is the place where Hephaestus had one of his forges.
👍 for the classical references.
I prefer Etna to Gainer. Some other "-er" suggestions
I also half thought of "Wailer" (also cf Bob Marley) but maybe that sounds too much like Wailing in agony.
Booster I get and don't dislike (but don't love either, like gainer). Spider and pumper I don't get. Can you explain? I also don't get Falcon, Torpedo or Erebus. If we go down the semi-opaque route, the idea would be to have something memorable without too much explanation.
@gkasprow Will have a think on some of the acronyms ;)
@hartytp Please look the preliminarny test results and if you like to comment or dicuss what would you like to achieve in the final performance of the RF PA. It is possible to adapt and little change the characteristics of the RF PA to your requirements.
Tomek, thanks for posting that! That looks really good!
Questions/comments:
Also, are you in a position to give me a more accurate cost estimate for these amplifiers yet? e.g. What would it cost if we bought 5 8-channel amplifiers (40 channels) or 10 amplifiers (80 channels)? If you'd rather not disclose that publicly then feel free to email it to me.
Thanks again for all the hard work on this.
PS @dhslichter @dtcallcock @jbqubit @jordens and anyone else who is considering buying one of these, do let me know if you have any other comments/suggestions. Or, if you have anything you'd like to see us test/check with the prototypes.
We hope to place a large order for these towards the end of January, so let me know if you want to join in and buy some at the same time.
Okay, name-wise, let's go for "Booster". Anyone who cares, feel free to express an opinion with 👍 or 👎 on this post.
@thomasRF I like your suggestion, but "Booster" is a simple name that I think physicists will find easier to remember, and it fits in better with the other names used in this hardware family.
Spider and pumper I don't get. Can you explain?
Doesn't need an explanation, needs to be reasonably memorable. If no good, just ignore. They are brainstorms :)
I also don't get Falcon, Torpedo or Erebus.
Again, just names. Erebus because it's a volcano and so I was joking around.
If we go down the semi-opaque route, the idea would be to have something memorable without too much explanation.
The whole point is no explanation, just something that people say "oh, that sounds like a name for a piece of equipment".
I am fine with Booster for now but if someone comes up with a better name I'm not wedded to it.
@thomasRF this looks very nice! I would say that we are definitely interested, based on these initial data. I have a few questions as well:
Edit: the gain shown in Fig. 3 looks to have a nontrivial deviation from linearity over a long range (from -20 dBm to -4 dBm input powers, for example). Is this an artifact or real? If real, can it be improved?
I would be interested to hear from @thomasRF on that, but I do believe this is real. I don't think it can be improved without significant reductions in power efficiency, which we wouldn't want to do for this design.
Remember that the main target for this design is driving AOMs and other things like that, which are far less linear than this amplifier design.
Other characterizations that would be interesting to have (@hartytp may already have discussed these):
Yes, those are on our list of things to test.
Is there data on how the amplifier handles poorly matched loads? How bad a mismatch can the amplifier tolerate on its output, especially at high power?
@thomasRF Your spec sheet says that the amplifier is unconditionally stable. Is that correct? What is the maximum reverse power it can take. i.e. what settings are you using for the reverse power interlock?
Edit: the gain shown in Fig. 3 looks to have a nontrivial deviation from linearity over a long range (from -20 dBm to -4 dBm input powers, for example). Is this an artifact or real? If real, can it be improved?
Anyway, as always, if there is an easy way of improving this without any significant trade-offs in terms of power efficiency etc then obviously we should do that. But, @thomasRF put a lot of work into the design and I do not believe that there are any real improvements to be had here without large sacrifices elsewhere, which is not worth it IMHO.
@gkasprow Can you add a few things to the schematic for the main board, please?
For the connectors that go to the RF modules:
Can you also add an overall power budget, please?
@hartytp ack to all.
@thomasRF some errata for the schematics:
Preamp protection:
Power stage:
Output protection:
@thomasRF Your spec sheet says that the amplifier is unconditionally stable. Is that correct? What is the maximum reverse power it can take. i.e. what settings are you using for the reverse power interlock?
If this is the Macom amplifier (NPTB00004A), the datasheet specifies no damage at a 15:1 VSWR, all angles (https://cdn.macom.com/datasheets/NPTB00004A.pdf).
If this is the Macom amplifier (NPTB00004A), the datasheet specifies no damage at a 15:1 VSWR, all angles (https://cdn.macom.com/datasheets/NPTB00004A.pdf).
Thanks for the reminder (10:1 was from memory). I'd guess that the loss in the output stage is enough to bring that close to unconditional stability (but @thomasRF will be able to confirm that).
@dhslichter Thanks for the other points. You can get all the part numbers from the Altium files (e.g. using the free Altium viewer). They are mainly correct IIRC. But, you're right that we should add annotation on the schematic so that they are visible in the printed/pdf version.
Could you either add those comments to the top post of this issue, or open another issue, please? Comments in the middle of the issue will probably get forgotten. Thanks!
Related point @thomasRF please can you check that all the part numbers in the Altium file are correct? For example, when I looked at the Atlium file before, it looked like the capacitance labelled on the schematic didn't always agree with the actual component part number when I opened up the component properties in Altium. This looks like a simple copy and paste error, but it's important to get it right (AFAICT, this will affect the BOM etc generated from the Altium files). @gkasprow Can you help Tomek go over the schematics and do some of this tidying up, please?
@hartytp will repost the above to a new issue. But the issues I have with part numbers are using the Altium files -- I am looking at the schematics in Altium and the part numbers listed on the .SchDoc files are not valid part numbers. For example, the schematic lists the power detectors as HMC7832, which doesn't exist. And the two preamps are MMA3128 and MMA1H22, which also don't exist.
Also @hartytp I didn't see any PDF schematics anywhere.
Also @hartytp I didn't see any PDF schematics anywhere.
Apologies, mis-remembered. Anyway, I was referring to the difference between looking at the annotations/labelling on the schematic file, and selecting an individual component, right-clicking and going into the "properties or whatever" menu. My recollection was that I easily found all the ICs when I looked at the part numbers in the properties tab.
@gkasprow after shipping the two channel prototype the channel cards worked themselves loose from the backplane (see pictures). It seems that the shelf holding the backplane as well as the PSUs worked its way loose and shifted about 10mm towards the back of the enclosure.
Simple solution would be to have end users install the modules themselves in the field.
@jbqubit that was my first thought, but the shelf holding the backplane and PSU will still move around in shipping, meaning the user has set this shelf back into position - this is non-trivial, as if it is positioned incorrectly the channel cards can have intermittent connections.
I noticed that during assembly. I will get rid of adjustable slots and make tight holes to prevent such movements.
@gkasprow great. I also noticed an intermittent problem that I cannot properly reproduce that I think was down to this same issue:
The second channel sometimes did not fully turn on (i.e. did not come out of standby). The channel card itself was fine, as if I swapped the position of the two channel cards in the enclosure both worked. The problematic channel card sits slightly lower than the other channel card, so I suspect the two prototype channel card enclosures are slightly different in size, and that the backplane was poorly positioned by me, leading to slightly intermittent connection on the D9.
@cjbe once we collect all issues, we will fix them in second chassis we have at WUT and ship it to you and you will ship yours to us.
@gkasprow Sounds good.
FWIW, I wondered if this would work better if the RF amplifier modules actually screwed into the D-type connectors on the control PCB.
We can drill 2 holes through top cover but module replacement would be more problematic.
Well, do whatever you think is best, so long as its robust and doesn't have intermittent connection issues when used in a "typical lab environment".
Sorry for my absence. I will try to answer in chronological order the question, starting from December 21st.
@hartytp
Apart from the electrical tests above, we should check mechanical things like: Are the fans easy to get to (these will probably need replacing at some point, so it's nice if they can be accessed/changed without too much pain)
- SMAs easy to plug and unplug
- Is field replacement of the amplifier modules reasonably easy?
- Are the SMAs robustly mounted?
Anything else?
I’m responsible for the RF PA modules design, and I discussed with Greg the issues of RF PA modules exchange and of the functionality of using RF modules. In the final version we will try to provide ease of replacement and usage convenience of the RF PA units. All comments, suggestions are welcome.
@thomasRF Good to hear from you, glad you're feeling better!
I’m responsible for the RF PA modules design, and I discussed with Greg the issues of RF PA modules exchange and of the functionality of using RF modules. In the final version we will try to provide ease of replacement and usage convenience of the RF PA units. All comments, suggestions are welcome.
Sounds good; this is more an issue for Greg than you, as you're focusing on the RF part, with Greg doing the mechanics etc.
We've finished our testing and posted some suggestions/comments about this. See the other issues.
@hartytp
- There are a few other tests we discussed for these amplifiers. Will you be able to do those in the New Year?
Yes, I ‘ll be able to do the discussed additional test, however, the only RF PA that I’ve made are in your possession. if you want, please send it (RF PA modules) along with the enclosure of the unit, which will be adjusted, and then I’ll carry out additional tests.
- Which amplifier design is this? Is it the Macom chip?
At the outset, I would like to point out that during the project I have made changes of the schematics diagrams, but the schematics diagrams at the GitHub diagrams are from a few months ago.
The RF PA amplifier has been designed in the three stages structure in two variants:
based on the NPA1003QA transistor you’ve proposed in the final stage, NPA1003QA: https://cdn.macom.com/datasheets/NPA1003QA.pdf
and in an additional option with NPTB00004A instead NPA1003QA (there is possible to achieve
slightly higher efficiency with slightly lower output power is possible.
NPTB00004A: https://cdn.macom.com/datasheets/NPTB00004A.pdf
The RF PA amplifiers that I have made, which you have, consists of NPA1003QA transistors in the final stage, and the two preamplifier stages are built based on the MAAM-010373 and MMG3012NT1 transistors respectively. The stages are design to work with NPA1003QA. In the additional option with NPTB00004A instead of NPA1003QA, if the NPTB00004A transistor should be used, is required to change matching circuits in the preamplifier.
The version that will be manufacturing in near future is depend on your choice. You are welcome to comment, and I’ll develop one optimal version for you.
The schematics diagrams I’ll update, after that we determine the final RF PA parameters, and you accept the test results so that I don’t need to generate subsequent versions of the documentation.
- Gain flatness is slightly worse than we originally agreed (we agreed +-1.5dB max). That's okay if it helps improve the design in other ways.
The presented RF PA solution is one of the versions possible to obtain with some small spread of parameters in the mass production. It is possible to change:
Please indicate what to do, what changes should be made (please remember that improving one parameter may worsen the other)
- Power efficiency looks great! 40% at 4.4W! Is that wall socket to RF, including SMPSs etc?
The power efficiency was estimated on the basis of RF input/output power measurements and power consumption at RF PA power terminals (i.e. for supply voltages of 8V and 28V). It does not take into account the SMPSs' efficiency. In addition, RF PA modules require power supply for protection, measurement and logic circuits, which are not power amplifier constituent circuits, however the additional blocks' power consumptions in relation to the amplifier are negligible.
- Is the abscissa in figure 4 supposed to be "Input power (dBm)"? (Not frequency)
Yes, the corrected report: preliminary report v1.1.pdf
- It is possible to adapt and little change the characteristics of the RF PA to your requirements." What kind of changes are you thinking of? Broadly speaking, that looks fine to me, but if you have any particular changes to recommend then do let me know.
As I presented in the answer to question no. 3, the presented RF PA solution is one of the versions possible to achieve with some small spread of parameters in mass production. It is possible to change:
- The cross-talk looks good here! Out of curiosity, do you have a feeling for what limits it? Is this limited by cross-talk on the power supply rails? Our of curiosity, are we confident that we're doing as much filtering etc as can be done easily to minimise the cross-talk? (e.g. the current number is fine, but if we can throw in a couple of extra ferrets/capacitors to improve it a bit then that would be great). Edit: you're probably already doing all this, but all power + data pins on those D-types need CMCs + CLC (or capacitor-ferrite-capacitor) filters on both the RF module and control PCB side.
In the project, I accepted as the maximum value of the expected cross-talk -60dB, taking into account the best is -80dB. Based on the preliminary analysis of measurements, I can say that the limitation in the first place is the RF PA circuits and the power supply circuits on the main board. Here, characterizing a single channel, I used a laboratory power supply and after that I connect it to the supply on the main board, and I observed increasing the noise flor at the output of the RF PA amplifier by 40dB , which is comparable to the RF PA amplification value.
For my part, I can make changes in the RFPA project to increase the cross-talk, so that in the full band can be much higher than 80dB. It is also required to improve spectral purity of the power supply (+8/+28V and negative voltage particulary) on the main board and the isolation of the power supply for the individual amplifier channels. In the part for which I am responsible, I can make these changes, so that in the next version of the amplifier, improve the inter-channel isolation, however, whether the overall effect will be satisfactory, depends on refilling RF PA modules and main board. I think that, I with Greg will deal with this issue. The question is what value of interchannel cross-talk would satisfy your needs.