Replacement for the PM9610 Prescaler for Philips PM6654 High precision Counter

[david69]

Thank You for Your idea.

[nctnico]

Yes, extra stitches are necessary and put an extra via (or two) at each capacitor in the RF path to connect the capacitor to the reference plane otherwise the current will flow along the edges of the copper ground pour on the top and the circuit won't work as expected.

[on4top]

someone has any empy pcb's available ?

regards

Tom

[david69]

I placed board with prescaler in counter. I have tested 50MHz
sine signal, counter starts rightly count from 45mV RMS and
logaritmic detector works fine, and disable prescaler without signal.
It needs some test on UHF, but I don`t have gener
The output from AD8313 can`t be pulled to ground

[david69]

test UHF 1GHz, works well, level sinus signal 100mV RMS.
AD8313 volume to ground occurs noice, 30mV RMS signal,
is good with MAR6, original PM9610 has 10mV level.
Chanel A date is 120MHz, but counts up 200MHz

[Fraser]

My 6654C needs its powers supply refurbishing due to capacitor ageing. It has been on my to-do list for ages so I am going to rehome it. This also means that I no longer need my Prescaler upgrade PCB's. After reading this thread I purchased two of the PCB's and some of the key components from the OP.

I will sell the PCB's and parts for what I paid for them. PM me if interested.

Update : Just checked on pricing......

I paid 30 Euros for 2 boards, 2 MMICs and 3 Prescaler-Chips. IIRC, at least one of the prescalers had been used by SaabFan for testing but carefully removed afterwards.

Fraser
UK

[SaabFAN]

I've now updated the board to revision F.

Changes:
- Attenuator upgraded to handle up to 12 Vrms, as stated on the front panel of the instrument, and dissipate 5,9 dB of that power.
- Prescaler-Footprints reconfigured with Jumpers, so different types of Prescalers can be used.
- Removed the Notch-Filter for the Power-Detector.
- Switched the connections of the OpAmp (Input is pulled BELOW 5V once RF-Power is applied).
- Negative power-supply of the OpAmp is now regulated with a zener to -3,7V. (Suggestion by a User of the Circuit Maker-community)

I'm thinking about using the unused board-space to add an AD8307 or a similar chip, capable of measuring RF-Power. An ADC or Microcontroller would then measure the output-voltage and make the digital information available by some means. USB-Connector at the rear or OLED-Display to be mounted externally would be possible.

[FransW]

If you have front panel connections, the rear BNC ones are not used.
Or the other way around.
The Philips mini-coax connectors are a bit of a mystery.
Either they are very obscure or a SMC type variant.
Checking around this week.
They seem to be used around the 80'ies / 90'ies.

Educational:  https://ewh.ieee.org/r1/new_york/tz/2007/20070315presentation.ppt
and from the same site: Quartz Crystal Resonators and Oscillators
Picture is from the same site.

Regards,
Frans

[SaabFAN]

Yes, the connector on the coax-cable connected to the BNC-Connector in the front-panel is strange.
But there's just enough coax-cable available for 2 attempts to solder an SMA-Connector to it. So I think using SMA-Connectors is the best way to go

[FransW]

Also for the +5V test connector?

[SaabFAN]

I've designed a new version. It still uses the MC12095 / 12093 Prescaler-Chips, but also adds a Power-Meter and an ADC to read the voltage coming from the chip measuring the RF-Power (about 36dB down from the actual input-power).

I have bought 5 boards and there are now 3 left, so if someone wants to get one of those, please PM me.

Attached are pictures of the new boards.

[FransW]

The original 1.5 GHz input board had it's input part shielded with a ni-plated brass box, soldered over the pre-scaler part.

Frans

[SaabFAN]

The shields on the original units were most likely there because Philips had to build the RF-Amp with discrete components with the technology available. For the first Version they even had to use through hole-components.

With the Amp implemented as a single MMIC, I don't think its necessary to put any shield on the board. Especially given the fact basically nothing else inside the counter is shielded


I have had some time to measure the highest possible frequency at -20 dBm this weekend: 2070 MHz.
The Signal-Source I used was my DIY Signal Generator, which uses a ADF4351 for anything higher than 400 MHz.
At lower frequencies (1.5 GHz), I can go as low as -29 dBm.
The lowest frequency the board was able to count was 8 MHz btw.

Higher amplification on the board might help with these performance-figures, but might also cause other signals to become too strong - Might be worth some investigation, but for the moment I'm pretty happy with the results

[SaabFAN]

I have updated the Boards to Mk. II Rev. A

Changes:
- Replaced T-Junction at output of RF-Amplifier with Resistive Splitter
- Added second RF-Amplifier to compensate for losses in Resistive Splitter and increase sensitivity
- Replaced the MC12095 Prescalers with MC100EL33 (MC10EL33, SY10EL33 and SY100EL33 fit also), as these parts are listed as "Active" by the manufacturers
- Replaced the NE5532 with a LM258 OpAmp in SOIC-Package to save space
- Schottky-Diodes of the automatic limiter-circuit replaced by BAT15-04 (2 Diodes in one package)
- Removed unnecessary Filters (High Pass at input of first Amplifier and Notch-Filter for the Detector)
- Replaced Zener-Diode for the negative supply with LM79L05
- New Board-Layout with larger and less Vias, as well as additional mounting-holes for a standoff (this standoff can be glued to the Mainboard, which makes mounting easier)
- Removed RF Power-Measurement features (caused problems)

The additional amplifier, together with the new Prescalers should improve sensitivity (hopefully not so much that noise is being counted ) and increase the measurement-range up to 2.5 GHz (currently the counter loses the signal at about 2.1 GHz but if I remember correctly, the normal inputs could count up to 160 MHz which gives a max Frequency of 160 x 16 = 2560 MHz).

Attached is a 3D-Rendering of the new board. I tried to avoid having to route stuff on the bottom side, but a few traces had to go down there. Undivided RF-Path is undisturbed though. I again went for a Coplanar waveguide-design but with larger Vias this time.

[geostep]

SaabFAN,

Thanks for the continuing effort improving your PM9610 replacement!  I'm looking forward to trying this out on my recently acquired PM6652.

- George

[SaabFAN]

Small update here to Rev. B (which I'm actually going to order):
- Coplanar Waveguide Vias now 0.5mm hole and placed in a denser pattern.
- Added a 0 Ohm resistor as a jumper in the +5V rail
- Replaced D4 and D7 with 1N4148 (before LL4148)

The addition of the jumper link and replacing the 2 SMD Diodes with Through hole basically adds a "3rd Layer" to have a completely uninterrupted GND-Plane underneath the RF-Traces (before the divided signal had to cross these 3 signals)

[SaabFAN]

Well, new boards are ready to order, but I since I discovered the opportunity to buy them ASSEMBLED from JLCPCB, I'm considering to do just that

Does anyone have any experience about buying them assembled from them? The Previewer sometimes shows (especially on larger chips) the chips to be in the wrong orientation. Wouldn't really make sense to have them assembled and then have to redo all the chips if they're in the wrong orientation.

[geostep]

SaabFAN,

Is it possible to ask them to run a test batch of one to see if all components are oriented correctly?
Perhaps with the understanding that you won't pay for their errors?

- George

[SaabFAN]

I went ahead and asked JLCPCB after making sure that my files are correct and they replied with the statement that their previewer has a bug and they have an engineer checking the rotation and correcting it if necessary.

BUT The parts I want to use aren't in their parts-catalogue and similar parts are "extended parts", which cost 3€ each to change the reels. So the boards will only be partially assembled (only 0603 and 0805 Capacitors and Resistors will be on the board) with "Some Assembly Required"

[geostep]

I can't wait to see what the partially assembled boards will look like when you get them!

- George

[SaabFAN]

New boards are here.

Assembly turned out to be quite good. Especially the 0402 Capacitors - Hard to solder those by hand and have it look good

[geostep]

The board looks great!  Really looks pro. And I agree, no matter how you try to solder 0402 parts (which are evil small), they never come out looking good.

I already PMed about the boards.

- George

[SaabFAN]

Btw. these are the parts that are missing and have to be hand soldered:

U1, U2: NXP BGA2869 RF Amplifier
U3: LM258T OpAmp (basically any dual OpAmp will work, but the ones with high bandwith might be prone to potentially damaging oscillations)
U4, U5: SY100EL33LZG-TR (Prescalers)
U6: LM79L05ACZ
D1: Bat15-04W
D2, D3: Bap50-03 PIN-Diodes
D4, D7: 1N4148 (Leaded diode)
D5: Bat15
L1: 680nH 0603-Package Inductor
VR1: Bourns 3296W-1-102LF (10k 10-Turn Pot)
C28: 47uF/10V Tantalum Capacitor - "C" or "D"-Package
R3: 352236RFT
R6, R7: 3522150RJT
P1: SMA-Connector (The ones that can be mounted at the edge of a PCB)
P2: 2,54mm Pin Header, 12 Pins, 90 Degree - I recommend ones that have longer pins, as the typical 90degree pin-headers with 6mm long connector-pins tend to be a bit too short. It is possible to solder a straight connector SMD-Style to the board, however. That way the pins should reach the contacts in the slot.
Jump_1: Any kind of wire that is flying around on the bench

The only ones that are a little tricky are the two MMICs U1 and U2. I recommend hot air or a magnifying glas and a small soldering tip for those.
Caps should be far enough away from the chips to avoid blowing them off the board if the airflow is not set too high.

[SaabFAN]

I just realized something: D2 and D3 must of course have their Cathode on the RF-Trace and the Anode connected to GND.
There's a small dot on the RF-Trace noting pin 1 of the diode, but it's a bit hard to see.

Unfortunately there was a screw up with the shipping labels where the chinese apparently forgot to put my name on the label. Just the street and the city. So I'm still waiting for the parts, hoping the chinese are able to straighten this out quickly with DHL. Otherwise I would be testing the prescalers right now.

Same issue happened with the boards btw., but they were shipped via UPS and it took one phone call with their customer service to clear it up. A day later I was able to pick up the boards at a UPS Access Point.
DHL is a bit more... "German" in that regard it seems (rigid and inflexible, with a fanatical adherence to protocol) and I had to mail LCSC and ask them to fix the problem

[geostep]

A big thanks to SaabFAN for sending me a pair of his new Philips prescaler boards. (Some assembly required!)

- George