Repair : TEMEX/SpectraTime LPFRS/LPRO 10MHz Rb generator [got GPSDO to cal]

[TheSteve]

Pic on eBay shows one with Trimble labelled OXCO. Will see what one exactly will arrive.

That one will work fine. I use the same gpsdo, I just mounted it in a box myself.

[dundee]

Cool. Are they same as mine?

Yes



Connected today the unlocked led as described in the manual. My led is not going off (lock). But the frequency is fine
on this device. Hmm...

[TiN]

How you know frequency is fine?

.

Hooked one of "good" Rb's to run last night. Attached RS232 data, samples every second.

[PA0PBZ]

Hooked one of "good" Rb's to run last night. Attached RS232 data, samples every second.

Did you notice that some parameters are reversed? Like photocell DC Voltage is 0 - 5V for FF to 00, and in your picture I see it start at the highest value, which is unlikely 
Also the heating currents are reversed.

[TiN]

Woops, u right.
Heating currents already correct. 2am here, my bad 

[MadTux]

Well guys, my rubidium standards are now finally back in Switzerland
And I have a few good ones too, didn't have time to test them all, though.
Good thing is that there is still plenty of rubidium left in the lamps. I disassembled one unit that didn't lock and looked at the lamp and there is a big blob of rubidium visible in the back .

It's easy to reassemble it back, you only have to push both the lamp and the rubidium cell back into that middle piece and solder the wires back in.

But before I do that I want to see whether I can reverse engineer the driver electronics and make the lamp run without the the rest of the assembly to take some pictures of that awesome rubidium glow. It looks really cool IRL (somewhat similar to potassium, which isn't surprising) unfortunately most cameras are unable to capture the full spectrum (looks far more purplish/red IRL). Tomorrow I'll compare phases of a few rb-standards over long time on my Tek scope to see how accurate they are. Maybe I can score a cheap 1PPS GPS receiver sometimes and use it as trigger source.

BTW, the coil for the inductively coupled rubidium plasma bulb runs at 147.4 MHz, with a few sidebands. (used a wire loop to pick up the magnetic field at the wires)

And that addon PCB inverts the lock signal: (wasn't already mentioned, so I thought I add it)
9 pin Sub-D pin 3:                                                 => 5V if locked, 0V if unlocked
10 pin rectangular connector on addon PCB: => 0V if locked, 5V if unlocked

Do I need a level shifter for RS-232 (from PC)? IIRC, RS232 from PC is +/-15V, while the LPFRS says it wants TTL (0/5V).

But now I have to go sleeping, it is already dawn outside

[SeanB]

Probably yes, a level shifter and inverter, as RS232 is negative as logic 1, and TTL versions are 5V as logic 1. Easy with a simple MAX232 chip, which will do both the generation of the 12V rails using some capacitors, and do the inversion as well. If you want to do it on a PC without RS232 ports use a cheap FDTI chip, which does TTL levels as standard. Just get the real chip, or fix the driver so it does not brick the clones.

[TiN]

it is already dawn outside

Sounds familiar.

You will need simple inverter and logic-shifter (5V->3.3V) if you use USB-UART like I do.

Thanks for teardown of phy module.
Do you plan to open cell section too?
I'm curious what that adjustment screw does near xtals.

[MadTux]

Got the lamp running. The complete oscillator is in the physics package, a couple of capacitors and BLV21 power VHF transistor behind the rubidium bulb. Schematics comes later, have to go to boring christmas lunch first.

Used a Schottky diode in TO220 for heating, damn efficient these things, the cables from the PSU got hotter than the diode  . A reverse protection diode might also be nice, because the HP E3616 sometimes went into overvoltage mode as the lamp ignited. Probably some reverse EMF when the magnetic field suddenly changes as the lamp ignites.

[TiN]

Here's schematics of LPRO adapter board, which is in place between RBXO in LPRO versions we talking about here..

[TheSteve]

Bought a 5 pack - so I'm in too!

[PA0PBZ]

Bought a 5 pack - so I'm in too!

I almost did, but decided to ask the seller about the shipping cost first. What did you guys pay?

[TheSteve]

Bought a 5 pack - so I'm in too!

I almost did, but decided to ask the seller about the shipping cost first. What did you guys pay?

Shipping to Canada was quoted at $110.00 USD so I shipped them to my US address which was 50 dollars.

[edpalmer42]

Shipping to Canada was quoted at $110.00 USD so I shipped them to my US address which was 50 dollars.

I don't know why he insists on using such a high-priced shipping method.  It makes most of his stuff uneconomical to me and I've told him so.

Ed

[TiN]

50$, DHL here. Good thing, it took only overnight to get it here (TW) last friday, so I could play on weekend. Sometimes it's worth it 

[TheSteve]

Received my 5 units today. Powered them up one at a time. The first 3 all seems to behave exactly the same. When powered up the frequency is about 200 Hz low. Then you can see the frequency go up but it never quite gets to 10 MHz, at the same time the current drops.

The 4th unit is similar but starts about 400 Hz low.

The 5th unit is actually generating a 10 MHz signal but it isn't as stable as it should be I don't think and the lock indicator never toggles.

Let the games begin.

Also received another GPSDO today, now I have three of them running - good thing they are cheap!

EDIT:

Just messing with one here.

If I just power one up it does what appears to be a bit of a half hearted effort to find 10 MHz but doesn't really cycle much. It just settles at 9.999823 MHz. However poking around with it open does cause it to do a proper cycle above and below 10 MHz - it even looks like there is a brief pause as it passes 10 MHz but it keeps going and eventually settles at 9.999823 MHz again. I believe the Peak voltage of Rb signal is too low and it never finds the proper dip it needs. The level I am seeing it only 0x20 or so which should be very close to the .6 volt minimum.

[Kleinstein]

The behavior if the first 3 units is about what one can expect from a unit that is not working anymore: The current goes down as the internal temperature gets close to staedy state - thats perfectly normal. The internal PLL tries to get lock, but never gets to the right frequency. Chances are that the internal VCXO has drifted too nuch to reach the right frequency. This might be relatively easy to fix by adjusting the coarce tuning so it starts closer to 10 MHz. 

[TheSteve]

What is odd about the units is that they "think" they are locked even though they are all off frequency. They are even all similar in that they think they are locked at 9.999823 MHz or so. If I adjust one of the pots I can make it ramp the frequency up and down forever, and it does pass 10 MHz. The range is from 9.999540 to 10.000120 MHz but it never locks/stops. Pretty easy to see it isn't centered around 10 MHz.

btw I tried the course electronic frequency correction but it is only good for +- a few Hz.

After the reading the manual I also directly connected the serial port to a USB to RS232 converter - yes technically it is out of spec but they recommend trying it in the manual. It works great and is nice and simple.

I will say that if/when someone figures out the problem with one it will fix probably 90% of the bad ones out there.

edit:

Just playing around, the GG values typically sits at 02 or so when it is hunting. When it finds a lock at 9.99982x MHz there is an increase in signal to around 20, sometimes higher - so it really is thinking it "sees" a valid Rb peak. I wonder if this much frequency error could be caused by a drive problem with the C Field drive coil - if it is weak perhaps the resonant frequency of the Rubidium spectral line separation has that much error. If we think about the division needed only a very small error at 6.8x GHz will cause a big problem at 10 MHz. My other thought is that the rubidium cells have become contaminated internally shifting the resonance point.

I also notice that the last value (AA) which is the 90 MHz power control signal seems too high on all of mine. It needs to be stable between 2 to 4.5 volts yet I am seeing values Dx to Fx - which would be at the high end of the range or in most cases out of range. Maybe the BLV21 transistor is weak - or that is the circuit that has drifted.

Fired up the spectrum analyzer - no signals detected around 147 MHz as someone else had posted.


thoughts???

[edpalmer42]

Generally, if an Rb standard thinks it's locked, it is!  Have you compared the frequency to one of your GPSDOs?  If you have a time-interval counter and two units that both seem to be off-frequency, use the counter to compare them.  If possible, collect the data electronically and process it with a program like TimeLab.  The difference between a pair of defective standards and a pair of working standards should be pretty dramatic.  You could also compare one Rb unit to your GPSDO.  Just remember that a GPSDO has it's own 'signature' that will appear in the data.

Here's a message that I posted a while ago that shows a comparison between two Rb standards.

https://www.eevblog.com/forum/beginners/help-me-understand-allan-deviation-measurements/msg801113/#msg801113

Ed

[TheSteve]

They are most certainly wrong - I have multiple GPSDO standards that I know are good. The Rb locks at over 180 Hz low which is a massive error.

[edpalmer42]

Frequency isn't the only thing to look at.  Regardless of the actual value, is the frequency stable?  Really stable with good Allan Deviation?  Then it is locked, regardless of the value.  Are you going to try to fix an RF problem when the fault is actually in the frequency steering or DDS circuitry?

You mentioned the effect of the frequency error between 6.834 GHz and 10 MHz, but you didn't get it quite right.  A 177 Hz error at 10 MHz would cause a frequency error of about 121 KHz at 6.834 GHz.  That's not going to happen.  There's no way that the unit would detect a signal that it thought was valid.

The C-field isn't the problem.  It provides a very small frequency adjustment and no matter how badly it was fried, it couldn't cause the measurement you're seeing.  Notice that the resolution of the C-field adjustment is shown in the spec sheet as 1e-11 per step as compared to the coarse frequency adjustment resolution of 1e-9 per step.

Your GG value of 20 is lower than the manual says it should be, but I don't know what that mean in terms of a circuit fault.  What are the values of the other parameters?

Ed

[TheSteve]

No the frequency is not as stable as it should be.(I don't think anyway) However the unit is finding something it likes, it will hunt up and down and find the same frequency each time.  The GG value also increases when it thinks it has found lock - the lock indicator often toggles as well. I was pondering the potential frequency error and at 6 GHz the error is huge as you say. I wasn't sure how much frequency change was possible via the C field but it makes sense it is a very small amount.

Here is a log from one of the units in question:
2B 0B 5D 44 80 D7 2B D3
31 0A 5C 4D 80 D4 30 E1
33 0A 5C 55 80 CA 35 D3
32 0E 5D 5C 80 CE 33 E4
34 0C 5D 64 80 D1 3D E3
37 10 5D 6A 80 C6 39 E2
3B 0E 5D 71 80 C1 45 D3
3D 0C 5D 77 80 C5 4F CF
3D 0A 5C 7D 80 C0 44 E1
3F 1F 5D 83 80 C9 52 E0
3F 2B 5D 82 80 C7 57 E1
44 25 5D 82 80 BC 5F D0
43 1F 5D 81 80 C2 60 E2
48 2B 5C 81 80 B1 5A DA
48 27 5D 81 80 C0 6A DB
4A 25 5D 81 80 B9 70 D5
4C 1F 5D 80 80 B6 74 D4
4E 27 5D 80 81 B5 78 D5
4F 2A 5D 80 81 B3 7C DB
52 24 5D 80 81 A7 80 D1

As the GG value increases from 1x to 2x is when the unit sweeps up in frequency and hits 9.999,817,1xx,x MHz at which point it indicates lock and stops there.

Just tested a second module, it stops at the same frequency, the lock indicator toggles and but eventually the GG value goes below 20 and it indicates unlocked again.

Third module shows the same behavior and frequency except it indicates it is locked and stays locked with a GG value in the 4x range.

-hmm, just had a small earthquake-

Forth module shows the same behavior and frequency, GG is under 20 so it indicated unlocked

5th module is does everything the same but it actually locks at 10,000.000.0xx.x MHz, it indicates locked but then the GG value falls below 20 and it indicates unlocked.

[Vgkid]

How is the power supply doing for those units? Have you monitored it.

[TheSteve]

How is the power supply doing for those units? Have you monitored it.

I haven't gone over it in great detail - would be nice to have a known working unit to compare to. My scope is currently in use monitoring my new GPSDO. I do see significant AC on some caps with my DMM. The 5 volt line looks good.

edit:

Did some quick bodge work and paralleled a few extra caps with the existing electrolytic units and now one of the units that was outputting 9.999,817 is magically giving me 10.000,000,0xx.x MHz and indicates it is locked. The Rb signal is still quite low though. The AC that was on the caps was up to 1 Vpp @ 125 kHz(internally used frequency).

So at this point, I'd say replace all 4 electrolytic caps and then see if they all work

[edpalmer42]

Here is a log from one of the units in question:
2B 0B 5D 44 80 D7 2B D3
31 0A 5C 4D 80 D4 30 E1
33 0A 5C 55 80 CA 35 D3
32 0E 5D 5C 80 CE 33 E4
34 0C 5D 64 80 D1 3D E3
37 10 5D 6A 80 C6 39 E2
3B 0E 5D 71 80 C1 45 D3
3D 0C 5D 77 80 C5 4F CF
3D 0A 5C 7D 80 C0 44 E1
3F 1F 5D 83 80 C9 52 E0
3F 2B 5D 82 80 C7 57 E1
44 25 5D 82 80 BC 5F D0
43 1F 5D 81 80 C2 60 E2
48 2B 5C 81 80 B1 5A DA
48 27 5D 81 80 C0 6A DB
4A 25 5D 81 80 B9 70 D5
4C 1F 5D 80 80 B6 74 D4
4E 27 5D 80 81 B5 78 D5
4F 2A 5D 80 81 B3 7C DB
52 24 5D 80 81 A7 80 D1

For those playing along at home, these values represent the outputs of the following parameters:

HH GG FF EE DD CC BB AA

HH (Photocell voltage) is low, but rising and nearing the acceptable range so the lamp might be okay.  As this voltage rises, I'd expect to see GG increase as well.

GG (Rb output signal) is low.  Just about anything could cause this.

FF apparently doesn't mean anything.

EE (VCXO control voltage) looks good.  It stabilized in the middle of its range.

DD (Analog Frequency Control Voltage) doesn't appear to be connected.  It looks like it's internally biased to the middle of its range.

CC (Rb lamp heating current) seems to be wandering a bit.  The manual says that it varies with ambient temperature.  I don't know if that level of variability makes sense or not.  Variations in the lamp temperature could affect the GG value.

BB (Rb cell heating current) looks like it was still warming up when this data was collected.  If it stabilizes, it should be okay.

AA (90 MHz power control) is moving from 4V1 to 4V5 which is at the top end of its acceptable range.

Since it looks like the unit was still warming up when these values were collected, I can't say that anything here is obviously wrong.  Even the GG values are trending in the right direction.  All I can suggest at this point is to let the unit run for an hour and see what the readings are and whether they are stable or wandering.  I'd expect them all to stabilize with no more than maybe a couple of points of jitter.  Over a long period (days or weeks) you may see a slow change in a few of them.

If you haven't already done so, try the C00 and F00 commands to return the center frequency to the factory default value.  I realize that your frequency error is far larger than these commands can set, but it might reset some other internal parameters.  Can't hurt!

Ed