Amateur Radio and the PCom equipment

AMATEUR RADIO USE OF P-COM EQUIPMENT

Technical data from various sources

IThis page is going to a collection of information from different sources. Once there is enough information to start sorting it all out I will make additional pages and add more links. If you have information about these units Send it to nr6ca at nr6ca dot org.

There is no organization yet so you need to page through the data. There is lots of good stuff here.

Steve and the group - I will make some comments about the PCOM ODU's. I am most familiar with the first generation which started to be surplused over 2 years ago. These were 23 and 39 ghz units with over 400 getting to the hams. I made the first contact with the scrap dealer.

Another 100 Remec units of a different design went to the LA area and are still being optimized. All work in pairs so there is a 21 ghz rx and 23 ghz tx on one end and the opposite on the other. the non-REMEC modules were made by Celeritek and have internal band-pass filters which need to be modified on the 21 ghz units. they can be opened, and with a microscope and diamond scribe, retuned, if you have steady hands.

The REMEC units were more broadband. The LO may be high side or low side, in the 9 to 13 ghz range. The synthesized LO's are too noisy for ssb, and several experts have tried to improve them.

The diplexer can be retuned in some cases, high side only.

ODU's which turned up here recently use an Endwave up/down conv, which is the old Millitech/TRW design. the bpf can be modified with a needle and I am just starting to characterize those.

I have not seen any of the ebay units up close. I know some 26 ghz units worked fine at 24, but i do not know about the 28 ghz ones. there was a 2nd and 3rd generation of 23 and 39 all integrated and very difficult to use. there was also a 49 ghz ODU. call me collect if you find any of those!

Good Luck, Will (W0EOM)

In an effort to get to 24 GHZ for both AO-40 and Terrestrial work, I recently came upon "ODU for parts" sale on ebay. I purchased 2 units and the 48 volt power supply and have done some research regarding these units. I also saw several other hams bidding on the units and I wonder if anyone here has more info than I can offer. NOTE: THESE ARE VERY DIFFERENT from the PCOM units that the SBMS has published a modification project.

The units are marked "TEL-LINK PMP 28 GHz Remote Tx 27.500-27.710 GHz, Rx 27.925- 28.135 GHz" . The PCOM web site indicates these are the remote units in a Point to Multipoint data network setup with a possible +22 dbm Tx output.

Interface to the unit is via a single coax cable that carries the 48 volts, the Tx IF, the Rx IF, and communication to a DSP engine on Card #1. The unit contains 8 plug-in Cards, 2 of which are the power supplies converting 48 volts to +12, +5, -5, and +7.2(for the 28 GHz module). Card #1 has the DSP engine and an FPGA. Card #2 has a shock mounted crystal oscillator. This card outputs 80 MHz (to card #4) and 1.85515 GHz (measured to Tx LO I/P on Card 6). Stability does not appear to be that great. Card #4 takes 80 MHz input and generates 3 output frequencies, 2280 MHz for Rx 1st IF, 285 MHz for Rx 2nd IF, 1995 MHz input to Card #3 for subsequent generation of the LO for the 28 GHz module. Of course Murphy lives here and on power up of the unit with Cards installed, a Tantalum Cap on Card #3 began to glow like a LED. The Lock Alm led is lit as are 2 other LED's and only the 80 MHz and 1.85515 GHz signals are measurable. All others are not operational. This may be due to a need for the unit to be attached to an indoor unit that sends data to the ODU for initialization.

While the spec sheet says the voltage input is minus, I found that the unit wanted +48 on the coax center pin in order for the ODU PS to work and have card 7 and 8 generate the +12, +5, -5, and +7.2 VDC ( without any cards plugged into the Motherboard). I have identified the voltage distribution to the 8 cards and connector for the 28 GHz module. I am in process of turning that into a PDF document if anyone needs it.

My first assumption is that most of the plug-in cards will need to be removed and a suitable LO added for use at 24 GHz. This, of course presumes that the diplexer can be retuned down that far.

Now the question, has anyone got more pertinent data about these units? Is it possible to tune the diplexer down to 24 GHz? Am I barking at the moon with this idea? All constructive comments welcome.

Steve K2IYQ

I purchased the PCOM ODU PS for 2 units. It is a 19 inch rack mount containing 2 switching 48 VDC PS, each having its own line cord and AC power switch on the back panel.

My units operate with the +48 VDC connected to the coax center pin.

The open ODU sitting on top is for reference to the PDF file I created. Arbitrarily I have labeled the cards from Right to Left as card 1 through 6. Card 7 and 8 are at the center of the photo (perpendicular to card 1-6) and contain the +48 VDC to +12, +5, -5, +7, and +3 VDC PS regulators.

*****WARNING***** If you remove any of the cards, they are NOT keyed and it is possible to reinstall incorrectly. The resultant release of smoke is very disheartening.

Steve, K2IYQ

As you may have discovered, the cards that plug onto the Motherboard in the ODU are NOT keyed. GREAT CARE needs to be exercised when re-plugging any of them. I have smoked at least 2 of the Tantalum Caps on card 3 and 4 due to plugging in just one pin off. I think I have replied to all the requests for my first cut at the motherboard layout and voltage identification. If I missed anyone, please drop me a note. Does anyone have any detail on the BSC Filter (a WDU 2729) ? I wasn't able to extract any detail from the BSC Web Site. If this is retunable to 24 GHz and the up/dwn converter will go down, there is hope for these units.

The 28 GHz Up/Dwn Converter looks to be built by SPC Electronics.

Steve, K2IYQ

If my conjecture is correct, these units do not look very good for amateur use. It appears an 8480 to 8550 signal is sent from the synthesizer board to the RX-LO port on the transceiver. This is first multiplied by three to the 25,440 to 25,650 range. These frequencies are then applied to the transmitter and receiver mixers. In the transmitter, the 2060 IF input causes the signal to be translated to the listed 27,500 to 27,710 output range and applied through the power amplifier to the output waveguide port.

For the receiver, this produces an IF frequency of 2485 for the frequency range of 27,925 to 28,135 listed in the specifications.

It appears the 2485 RX-IF is run through the 2485 BPF on Board 5 then downconverted by the 2280 1st local oscillator input frequency to 205 MHz. The 205 MHz signal is then upconverted to 490 MHz by the 285 MHz 2nd LO input signal, and run through the 490 BPF before being applied to the band separation filter on the mother board.

So the center frequency of the received information on the cable appears to be 490 MHz while that of the transmitter signal is 205 MHz. The fact the receiver IF is first converted to 205 MHz is an interesting coincidence. It is likely the 205 MHz stage is used for either clock recovery, or for purposes of phase locking the crystal oscillator to the base station transmitter's frequency. Interesting.

It appears the only interboard frequency that gets varied as the operating channels are changed is the 8480 to 8550 output of synthesizer card 3 that is applied to the LO port of the transceiver. Interestingly, this LO signal serves both transmitter and receiver.

If I am correct in my suppositions, it will take some work to convert the transceiver units. The input/output bandpass filter unit is probably beyond conversion to amateur use as the WR-28 waveguide is approaching its cutoff frequency when the frequency is reduced to 24 GHZ. The wavelength in the guide increases to about 1.6 times that at 28 GHz. I would think this would play havoc with the length of the sections.

Yet, this filter is probably critical in eliminating the transmitter and receiver image frequencv and response created by the mixers in the block. Some form of filtering will likely be needed. I would guess that there is some filtering on the transmitter side built into the block so that the transmitter PA does not have to simultaneously amplify both mixer products. But, if so, this will have to be retuned for the 24 GHz range too.

Bruce KG6OJI

I spent an hour or so taking a closer look at my P-Com unit. It turns out the transmitter card (Card 6) saturates with an input of about -30 dbm at 205 MHz applied to the N connector. Card 6 apparently simply mixes the applied TX-IF signal with the 1855 input signal and filters out the upper sideband of the mixer products. The on-board filter seems to me marked 2060 MHz so that makes sense.

Checking the overall response from the N-connector input to the TX-IF O/P (I assume O/P stands for output port?) connector, I find the peak of the passband occurs around 2060 MHz with the -3 db points at 2026 and 2103 MHz. The maximum output signal at the jack is -7.1 dbm at 2080 MHz with -30 dbm at 230 MHz applied to the N-connector. The power of the incomming 1855 signal is 0 dbm.

The LO port of card is outputting 8480 MHz at +14.8 dbm. THere is no chanfe in this signal as a function of changing the TX-IF signal at the N-connector.

I could not measure any output, whatsoever, at the TX waveguide port. I could not even measure any local oscillator leakage at the receiver waveguide. This with an H-P 432A power meter set to the -20 dbm scale.

Again, to get a signal from Card 6 I had to connect a 100 ohm resistor from the TX PWR-SET pin to +5 Volts.

The most enigmatic finding is that I do not have +7.2 volts on the transmitter connector as you show. I explored this a bit with an ohmmeter and found there is no path between this pin (these pins) of the tranceiver card and the power supply module that you designayte as generating the 7.2 volt signal. I thought this might be why the transmitter is inoperative. So I unpacted my second unit and found the same condition -- no 7.2 volts on Pins 1 & 9! Now I am really puzzeled. It may be there is a switching circuit you activated when you applied the full 48 volts to the unit suddenly. I have been able to run mine up to 40 volts. Everything else appears normal.

Checking the transceiver power cable I found the connector numbers match exactly pin-for-pin. Pins 1 & 9 are tied together in the unit, apparently as you indicate for 7.2 volts. The utilization of two pins as they do for 5V and ground suggests Pins 1 & 9 constitute a major supply circuit. Pin 2 is -5 volts as you indicate. Pins 3 & 7 are +5V, Pins 4 & 8 are ground, Pin 6 is +3.5 volts. That leaves only Pin 5 as a possible control signal. This pin measures 25 ohms with a positive voltage applied and 16 ohms negative. It is quite possible that it is a circuit that samples the output peak power level and sends a "stretched" and integrated proportional signal to the amplitude controller to keep the PA out of saturation that would cause severe intermodulation. Anyway, I am now thinking it is the lack of the 7.2 volts that is muting my transmitters.

Bruce KG6OJI

This morning I played with the numbers and the information you dug out about the P-Com units, and think I may have come up with the general frequency conversion architecture. I made a sketch; however, I do not have a scanner available. If you have a FAX machine available, I will send it. If my conjecture is correct, these units do not look very good for amateur use. It appears an 8480 to 8550 signal is sent from the synthesizer board to the RX-LO port on the transceiver. This is first multiplied by three to the 25,440 to 25,650 range. These frequencies are then applied to the transmitter and receiver mixers. In the transmitter, the 2060 IF input causes the signal to be translated to the listed 27,500 to 27,710 output range and applied through the power amplifier to the output waveguide port.

For the receiver, this produces an IF frequency of 2485 for the frequency range of 27,925 to 28,135 listed in the specifications. It appears the 2485 RX-IF is run through the 2485 BPF on Board 5 then downconverted by the 2280 1st local oscillator input frequency to 205 MHz. The 205 MHz signal is then upconverted to 490 MHz by the 285 MHz 2nd LO input signal, and run through the 490 BPF before being applied to the band separation filter on the mother board.

Regards, Bruce KG6OJI

... don't give up on the PCOM units yet. All seen to date use a common LO for tx and rx, with the IF's determinining the channels.

On the earlier units, all that was usuable was the upconv, down conv, and the isolators. the large copper diplexer was not tunable to 24.2 ghz except for a few 26 ghz ODU's.

The synth LO was too noisy for ssb on 24 . all the rf units have built in doublers and also internal bpf's, which is a critical item. some can be returned.

As soon as i get one of the rf modules I will open it up.

I am right now looking at some 21/23 ghz Endwave modules which are separate T and R.

Milliwave made a lot of the 39 ghz triplers found several years ago. They were bought by TRW, then merged with Endgate to form Endwave. time marches on.

Will Jensby

We have not given up, just doing more looking and testing. I am updating the PDF drawings and will issue new ones in a day or two. Bruce has drawn a block diagram that is a POSSIBLE block of the units. His math seems to work. So as soon as I get that in usable electronic form, I will forward it to the group.

In the mean time, does anyone have the ability to measure the phase noise on the LO signal? That would be helpful to know. Second, does anyone have the ability to determine the frequency of the crystal in the unit? The crystal may be the reference for all the VCO's. Third, there is at least one PIC 16F84 micro processor on the Synthesizer card with a connector near it that might be for loading/changing it. Can anyone make that determination?

73, Steve

Dear Fellow P-Com "Parts Units" Analysts,

From the lack of recent postings, one might assume interest in these units is waning. At the risk of boring everyone, here are some brief findings.

I unfolded the crystal oscillator deck for examination and found it is connected to Card 4 with three, 3-conductor ribbon cables. All three conductors of the cable off the bottom of the board are used for grounding. The center pin of the side cable is grounded. The top pin is marked TV tuning voltage). The bottom pin is marked RF/+12V. The output at the latter pin is 80.006+ MHz. I believe the crystal is oscillating at 80 MHz, but that's only a guess as I could not find any frequency markings on the exposed surfaces.

Probing around the nearby portion of Card 4 with a 10:1 scope plug connected to a counter, I found 80 MHz, 40 MHz, 20 MHz, 10 MHz, and 5 MHZ... that is approximately these frequencies. The oscillator is running at about 80.006 + MHz. From the LED on the other side of Card 4 that remains lit, I assume the oscillator is running unlocked. During normal operation, the oscillator might get locked to the base station frequency (perhaps via the 205 MHz receiver IF frequency).

Card 4 is marked "P-COM IF-LO-VCXO". A terminal near the upper edge of the crystal oscillator board is labeled "VCXO TV". The 80 MHZ output port measured 80.005978 MHz with a counter; however, when a 20,000 ohms-per-volt meter on the 10 V scale was touched to this point, the voltage read +8.5V and the frequency dropped to 80.005945 MHz. Connecting a 1000 ohm resistor from this point to ground with (the meter connected) dropped the voltage nearly to zero, but had little further effect in diminishing the frequency. Perhaps, when the system is in operation and locked, the oscillator moves to exactly 80.000 MHz.

I still have not been able to obtain any transmitter output from the transceiver module. I opened the transceiver unit and found it has a rather complex control board with an RF absorber cemented to the bottom -- obscuring the printed wiring. The RF circuitry is readily visible and straightforward. I can FAX a crude sketch if anyone is interested.

Bruce KG6OJI