TNP #40 - Stanford Research FS735 Dual Distribution Amplifier Teardown, Repair & Experiments

In this episode Shahriar repairs a fault SRS FS735 dual distribution amplifier. These instruments provide two independent 7-channel 10MHz distribution modules with exceptional signal purity and isolation. The instrument does not properly powers on which points to a potential problem with the internal power supply.
The unit has two identical power supplies; one for each module. Both power supplies have a failed startup capacitor which can be easily replaced. After the repair, the power supplies are verified for functionality and power delivery. The schematic of the entire unit is described as well as a view of the module PCB. The performance of the unit across frequency, harmonic distortion as well as isolation behavior is also measured and verified.
Please consider supporting The Signal Path:
www.TheSignalPath.com/Donate
/ thesignalpath
Other links:
www.TheSignalPath.com
/ thesignalpath

[@jhonbus]

Ahh, the classic "ovenized bootstrap capacitor" PSU design...

[@bioazard50]

Had the same failure in our lab. I ended up replacing both power supplies and it's been running perfectly for a few years!

[@trevorvanbremen4718]

Thank you VERY much Shariar! While I do love your typical video where you play at the 'daylight' end of the spectrum (GHz / THz), I truly appreciate that you still throw out videos down at my DC end of the spectrum (MHz in this case)

[@purerhodium]

I love Pooch.

[@IanScottJohnston]

Those 4mm banana adaptors on your BK Electronic Load are worth their weight in gold.....nice one!

[@vincei4252]

Do you know where to get those? I have the same electronic load and have considered going to the trouble of making my own but if I can buy some from somewhere all the better.

[@IanScottJohnston]

@@vincei4252 Ultra One Design (all one word) on T_indi_e (a bit cryptic coz YT is deleting my post thinking it's spam)

[@vincei4252]

If someone replied I can see that KZbin has deleted one of the comments because of course it has.

[@garci66]

Loved the Chernobyl reference ;).

[@ChrisSmith-tc4df]

I have a pair of the half-sized FS735 10MHz distribution amps. I found the limiting/reshaping functionality quite novel and unexpected, but that does ensure a consistent output amplitude over a wide range of inputs without relying upon AGC as well as transforming anything in to sine wave output.

[@R2AUK]

👍 The schematic doesn't look complicated and doesn't seem to contain rare parts. Looks like something that can be easily replicated if necessary.

[@ChrisSmith-tc4df]

The only critical part might be the tunable IF transformer as they’re typically made for 10.7MHz, and it needs to be able to be detuned down to 10MHz.

[@WAEviltechie]

Distribution amplifiers are very frequently used in the broadcast industry for SDI, but also for analog reference signals like black burst or tri-level sync. Could be fun to take a look at one of those and see how it compares.

[@SeanBZA]

Yes, have one, with plug in amplifiers made by Opamp Labs, Had the one fail, so had to fix it, as getting a replacement would take months. Removed the can, and then soaked the potting in epoxy remover for a week, to make it soft enough to pick off from the components. Turns out the unit is a bog standard LM709 with a booster stage on the output. Fault was one of the diodes in the booster stage went open circuit, and thus both transistors went open circuit in sympathy. So in went a 2N2219 and a 2N2907, with push on heatsinks, and a pair of salvaged silicon diodes, as they had a lower forward drop than the 1N4148's I had around, and this thing has no emitter resistors. 3 new tantalum bead capacitors on the power input, as the existing colour coded ones had sort of disintegrated during potting removal, and it was working again to send video through the system. Gain of 2 only, so it could run up to 10MHz with little problem, though I would guess it dropped off badly past that, but for video and audio it was perfect. Odd design, everything is plug in, 4 plug in amplifiers that are identical, and then another potted plug in power supply that gives 15V rails for them. Then input is via BNC, with a 75R terminating resistor, and output is simply a 75R resistor soldered to the BNC inner, and then a wire linking them to the output coax, shield connected only at the opamp side. Going to guess output isolation is not great, but it worked well for the application it was used for. modified the 2 parts of 4, simply by snipping the 75R resistor, to make it run for stereo audio, which it was great at. Video not so good, driving 15R of load was not great, thus the fixed one needing heatsinks to keep it cool. Still better than the Matsushita amplifier it replaced, which used printed carbon resistors, and which was from the 1970's, with early silicon transistors that tended to be noisy. That one went to CCTV use, where the camera noise was larger, so it was not visible on the small CCTV monitors.

[@Richardincancale]

This video reminded me of a video that I now see is ten years ago(!) by Gerry Sweeney on YT on how to take an old Rubidium clock from eBay and turn it into a 10 MHz lab standard. IIRC he used a cable TV multi-port amplifier to provide the functionality of the SR unit here. It would be fascinating to compare the results!! In practice it worked fine for his home lab, maybe not to Shariah’s standard though :-)

[@gwyllymsuter4551]

Always informative

[@ChrisSmith-tc4df]

I would have used the BNC T at the scope input such that it is the shortest stub on the 50ohm transmission line.

[@Thesignalpath]

Yes, you are right. That would have been better.

[@ChrisSmith-tc4df]

⁠@@ThesignalpathI wonder if you had power divider on your mind as you mentioned your desire to potentially perform a additive phase noise measurement, and typically a resistive power divider is where the T was. Did the R&S input impedance limitation punk you mid-build?

[@aqib2000]

Hi and welcome the noise path

[@blays.7178]

Just went on google to see if i could afford that lcr meter and hoooooooly meter, this thing is way up to 15k.. I guess i will use my f generator and scope for a little more time.... But maan this thing can measure leakage and everything... Could not imagine what else could do in L mode... you are a lucky guy to play with tools like that and you also posses knowledge that is very very rare... i watch your channel from very beginning and i love it every single minute... You are super professional and know your stuff... Thank you soo much, to share that knowledge like that, you truly are awesome person :] also say hi to pooch :D

[@raymiller5738]

Thanks, Shahriar, seems to be a theme going with SMPS, looks like some component specification issues throughout the industry. It is possible to get to the bottom of some of the common failure modes?

[@PP-xy9bg]

I’m gonna talk to Rhode about that :) Please, don’t hurt them 😅

[@lasersimonjohnson]

Wish I could add a second like for the feline content. 😅

[@thomasleerriem6872]

Failing switching power supplies... what a surprise :D Companies don't like linear power supplies anymore, because they last forever.

[@vincei4252]

The ex-girlfriend joke made me laugh more than I should have 😂

[@anandarochisha]

Glad it's not an ex- wife..

[@muhdiversity7409]

@@anandarochishaYeah, those can be devastating.

[@AndrewZonenberg]

Is this essentially the output stage of a FS752, but without the GPSDO driving them (requiring the 10 MHz to come from somewhere external)? I've been dreaming of getting a FS752 for my lab for a while now.

[@Bllinker]

I wonder what the collector resistors in emitter followers are for (R212, R215, R217 for the first channel) - not the RC filters, the ones after them. Back in uni we were taught that they're sometimes used for decreasing thermal dissipation in the transistor itself, but maybe here they're there for some additional filtering of the harmonics (increased input capacitance both because of the decrease of quiescent collector-base junction voltage and the Miller effect)? Maybe not, maybe it has to do with decreasing buffer "transparency" somehow? Or maybe it's some additional power supply separation, like a second stage RC filter without the C, since a "bumpy" VCC doesn't make much difference for an emitter follower? Yeah, seeing as those collector resistors only drop around 0.25/0.5/1.0V respectively, I'm going to go with that last explanation.

[@klebs6]

waa waa wee waa

[@andrewdignan3038]

Hi Shariar How about a tutorial on practical measurement of return loss in rigid or semi flexible waveguide using the fieldfox or similar type instrument. Many thanks. Andrew

[@OneBiOzZ]

very nice video, some curious decisions. Not sure why one would want a dual distribution amplifier instead of one big one. I have never needed two independent clocks, although i understand it as far as construction and options go, its just annoying to have a T and two cables to feed it. Also im not sure why they did not length match the feed for the output amplifiers. Yeah it might be half a ns or so of delay but it costs very little and you can increase your channel mismatch specs.

[@SeanBZA]

2 amplifiers as if you have a lot of equipment some might need a 10MHZ signal, and then you might want to have some fed from the GPS directly, and then all the rest fed from a GPS locked Cesium source, so that you have a long term stable signal, and a separate short term stable one, as the the GPS disciplined oscillator removes all the phase jitter off the GPS signal.

[@lukahierl9857]

Prrhaps for phase noise mesurements with cross corelation?

[@johncundiss9098]

@@SeanBZA Why?

[@johncundiss9098]

@@lukahierl9857 Why? If you want to measure that just do it directly.

[@johncundiss9098]

They did not match the length of the input line to the amplifiers/filters ether. I agree with your words. When it comes to metrology grade, everything matters.

[@foobarables]

What might be the reason for placing to ceramic caps into parallel at the input? They are also on the PCB, C109 and C110.

[@Thesignalpath]

Part of the filters.

[@ChrisJackson-js8rd]

what is the advantage of measuring delay in nanoseconds instead of phase shift in degrees?

[@Thesignalpath]

No particular advantage in this case. It is just in absolute time which is independent of frequency.

[@antoineroquentin2297]

Is there a groundplane on the backside of the PCB or is that not important at that frequency?

[@Thesignalpath]

Yes, there are ground plans.

[@der.Schtefan]

"Ah, that's where I placed my ex". 😂😂😂😂😂

[@KallePihlajasaari]

If you distribute 10MHz sinusoid to all your frequency instruments on the benches with such an amplifier how much QRM will there be in the lab at 10MHz. Will is show up on a short piece of wire stuck into a spectrum analyser above the noise floor?

[@TheBananaPlug]

The schematic shows the input transformer as being 2:1, per the legend, but the part (also on the legend) is a TC4-1T which if I recall is a 4:1 transformer form mini circuits. Have I confused my part numbers or is there an error on the board?

[@MrJrwillison]

A 2:1 turns ratio provides a 4:1 impedance ratio (which is how Minicircuits specs, and numbers, its transformers)

[@TheBananaPlug]

Thanks, makes sense@@MrJrwillison

[@vmiguel1988]

I think for something you want to run 24h you should had replaced all the caps since one already failed means the others are not that reliable anymore!

[@Thesignalpath]

I did check the rest of the capacitors in circuit and they are still doing well. :)

[@kirkpennock2997]

Startup cap goes flat. Why?

[@pa1wbu]

Cooked dry by a nearby heatsink is a typical scenario for an electrolytic capacitor

[@dtiydr]

2:33 My stomach hurt so bad..

[@georgkrahl56]

I do not see the point of stabilizing the emitter current of the differential amplifier. Doing so the transconductance and therewith the amplification of the differential amplifier will decrease with increasing temperature. Usually this is compensated by increasing the emitter current with temperature, simply achieved by not taking these compensation measures. See eg the CA3028 IC. But maybe the exact amount of the amplifier is of minor importance, it operates anyway in clipping mode.

[@neonkev7866]

Haha that ex-girlfriend crack caught me off guard! Of course, it was just another dead capacitor. If only you could replace other things as easily...

[@neonkev7866]

@alifesh If she was really incendiary, she's a tantalum.

[@Rob2]

It would be even better when one half was for distribution of 10 MHz and the other half for 1PPS (1Hz). But they probably make another model for that.

[@MrJrwillison]

Indeed we do. (Just Google SRS FS735.) The FS735 can be ordered with any two of the following: Option 1 - 10 MHz distribution amplifier (two are shown in the unit in this video) Option 2 - 5 MHz distribution amplifier Option 3 - CMOS logic distribution amplifier (just right for 1pps distribution) Option 4 - Broadband 50 Ω distribution amplifier (works for video, dc-100 MHz) Option 5 - Broadband 75 Ω distribution amplifier

[@johnshaw359]

Now I see why the PSU's were banished to the far-side of the board.

[@AdityaMehendale]

:o Ex-girlfriend-joke? That's a first for Shahriar - no?

[@loueckert4970]

Reverse engineering at its finest... 73 de KT1R

[@deadbugengineering3330]

The volume of the screeching might increase a couple dBs if she watches this video. 😅

[@wolpumba4099]

*Summary* *Introduction and Overview of the Stanford Research Model FS 735* - 0:07 Introduction to the video, discussing the Stanford Research model FS 735, a dual distribution amplifier. - 0:17 Explanation of what a distribution amplifier is and its use in synchronizing multiple equipment in a lab. - 0:37 Explanation of how a distribution amplifier works with multiple instruments, preventing phase noise and drift. - 0:56 Introduction to the specific features of the Stanford model, including independent distribution amplifiers and seven outputs. - 1:25 Close look at the back of the device, showing the 10 MHz input and seven outputs. *Device Issues and Initial Analysis* - 2:00 Explanation of the additional feature of the device that filters and perfects the sinusoid output. - 2:24 Revealing that the device is currently broken and needs repair. - 2:44 Examination of the inside of the device, highlighting the two fully independent amplifiers and power supplies. *Power Supply Issues and Repairs* - 3:24 Analysis of the power supplies and possible issues with capacitors. - 3:59 Detailed examination of the power supply architecture. - 4:42 Decision to remove and measure the capacitors to identify any issues. - 5:33 Test of the power supplies after capacitor replacement. *Detailed Examination of Circuitry and Schematic* - 6:38 Examination of the circuitry handling 10 MHz conditioning. - 7:01 Overview of the schematic diagram of the amplifier. - 10:50 Review of the printed circuit board. *Testing and Final Observations* - 12:59 Measurement of the frequency response of the distribution amplifier. - 14:26 Changing the input to a square wave and observing the output. - 16:56 Checking the insensitivity of the instrument when some ports are improperly terminated. *Conclusion* - 18:03 Conclusion of the review and repair of the Stanford Research model FS 735 dual distribution amplifier. Disclaimer: I utilized gpt4-32k 0613 to condense the video transcript into a summary. I employed Prompt 1 to generate timestamped bullet lists. I then used Prompt 2 to organize these lists into sections with titles. The summary was manually formatted using KZbin comment markup. Prompt 1: "Generate a bullet list summary, including starting timestamps for each point." Prompt 2: "Split the following bullet list into sections. Create section titles. Keep timestamps for the bullets. Use this format for titles: *title*."