Kontron/Tabor 6010 universal counter still worth getting?

[0xdeadbeef]

I'm looking for a universal counter but don't want to spend too much money (ideally <200€). It should have at least a TCXO time base with like <= 1ppm but also allow to use an external 10MHz reference.
Doesn't have to support GHz stuff, I'll mainly use it to check drift of 10MHz time bases like OCXO vs. GPSDO and things like this. So anything <= 100MHz will certainly do. And well, the smaller the better. I already have too much gear and not enough space.

So I first looked into Racal/Dana counters like 199x but what I found was still rather expensive and in pretty bad condition and for counters from big brands like HP/Agilent it's even worse.
That's why I currently think about getting a moderately prized Kontron/Tabor 6010 with a TCXO time base. I assume the 60x0 counters were produced from the late 80s to the late 90s or so. Maybe even longer.
My understanding is that they were built by Kontron first, then that business was sold to Tabor or maybe Tabor produced them as OEM or whatever. Anyway, the seem to have been produced for quite some time and you can still download datsheets, drivers and manuals for them from the TaborElec web site.

Anyway, I'm unsure if it's really worth getting such a counter which will be at least 20 years old, maybe even older. Like there's the chance of dying the death of dried electrolytics, the TCXO probably drifted quite a few ppm over the decades, the calibration is fully manual and I'd need to get some GPIB to USB adapter if I ever want to control it from a PC. Actually I figure I need to open the 6010 to calibrate it and to configure stuff like the 10MHz reference output to input.

So in a nutshell, is it still worth getting something like this for like <= 200€ or should I rather get my hands on something new like a TTI TF930 with USB and electronic calibration but costs more like 420€ (incl. VAT)?
To be perfectly clear: I don't consider buying an old counter because I'm into vintage stuff. It's more like I hope to get good gear for a good price in this case.

[EDIT]
The Tabor 60x0 was obviously also also sold under names like Keithley 776 and Racal 2202.

[perdrix]

I'd say so - they are good counters, were badge engineered as Racal in the 2022(R)!

Dave

[don.r]

You might also want to consider older HP gear like the 5315/5316. The 5316 has a 10Mhz external input. These can be found for under $100 frequently.

[jpb]

It might be worth looking out for a second hand TTi TF930. These are simple reciprical counters with no interpolation but do produce continuous counts and have a reference input and a computer interface via usb. They can do 10 digits on 100s or if you use the computer interface you can collect data continuously over longer gates.

One drawback is they have no backlight on the display.

New price is a bit excessive but they seem to be sold second hand for under £100 from time to time. There is also the TTi TF960 which is more recent.

The other thing to say is that the reference input is a bit deaf - it needs TTL or 1Vrms AC and a lot of sources are less than this which is a bit annoying.

[0xdeadbeef]

Yeah, I mentioned the TF930 in my initial posting. Problem is that I couldn't find anything cheap yet. Actually the few ones sold on eBay are much more expensive than a new unit from RS (including VAT) which is around 420€ and a bit more than I'd like to spend.
Besides, I don't think I like the external reference input on the front and I guess some thinks like Phase A/B are missing (which I don't really need though) compared to the 6010.
Then there's the issue that there are two inputs, but the B input is only for >80MHz. So I couldn't even use stuff like Ratio A/B with a 10MHz OCXO vs. GPSDO.
Again, not a total showstopper. It's just that this makes the 2nd channel more or less useless for me.
The only measurements that the TF930 has but the 6010 doesn't are pulse width and duty cycle measurement. Dunno how much sense these make though as they depend on rise/fall times and threshold.

About the old HPs: my impression is that these are at least 10 years older than the Tabor 6010 and the ones that are in the <200€ range don't seem to have a TCXO.
Anyway my chances to get a mint unit for a good price are better for a less known brand like Tabor.

[jpb]

Yeah, I mentioned the TF930 in my initial posting. Problem is that I couldn't find anything cheap yet. Actually the few ones sold on eBay are much more expensive than a new unit from RS (including VAT) which is around 420€ and a bit more than I'd like to spend.
Besides, I don't think I like the external reference input on the front and I guess some thinks like Phase A/B are missing (which I don't really need though) compared to the 6010.
Then there's the issue that there are two inputs, but the B input is only for >80MHz. So I couldn't even use stuff like Ratio A/B with a 10MHz OCXO vs. GPSDO.
Again, not a total showstopper. It's just that this makes the 2nd channel more or less useless for me.
The only measurements that the TF930 has but the 6010 doesn't are pulse width and duty cycle measurement. Dunno how much sense these make though as they depend on rise/fall times and threshold.

About the old HPs: my impression is that these are at least 10 years older than the Tabor 6010 and the ones that are in the <200€ range don't seem to have a TCXO.
Anyway my chances to get a mint unit for a good price are better for a less known brand like Tabor.

You make some valid points.
If you are looking at 10MHz ratio then you just use one as the reference - I used to set my 10MHz OCXO by using it as a reference and measuring the 1pps from a GPS averaged over 100 seconds - one nice thing is the lowest frequency is 1mHz so measuring 1Hz is fine - some other counters start well above 1Hz.
I agree though that you'd be wasting the second channel.
I generally don't use  the second channel but recently I bought an RF generator on ebay and was glad to be able to go up to 3GHz (I have a much more expensive Tektronix Counter FC3100 but that only goes up to 300MHz).

I bought my TF930 from TTi themselves as an ex-demo model, it cost me £190 I think.

I am not particularly pushing the TF930 though, it definitely has its flaws such as the non-backlit display, the ref port on the front (as you say) and the fact that you need a high level (3V+ TTL or 1Vrms sine). Also it is a bit lightweight with plastic rivets rather than screws. But it also has some good points - very light, draws little current so can be powered from its internal batteries or from the USB port and the fact that it has no dead-time and is a proper reciprical counter.

[0xdeadbeef]

Yeah, for <= 200€ I'd really consider buying one but I don't see this happen at the moment.
Side note: some sellers seem to show a version with different front layout (older version???)
https://www.ebay.de/itm/AIM-TTI-INSTRUMENTS-TF930-FREQUENCY-COUNTER-TIMER-3-GHZ/271964144601?epid=1026587164&hash=item3f525383d9:g:X2UAAMXQatBSa78h

And while we're at it: while the TaborElec site and the datasheet speaks of a "Standard TCXO stability Oscillator", the manual than can be downloaded there clearly describes the TCXO as Option 1.
So at least older versions of this counter seem to have a standard oscillator (5ppm) that is not a TCXO. Actually the device shows all installed options at power up after the display test.
It should display "6010-1" for TXCO or "6010-1.2.4" for TCXO, 1.3GHz input and GPIB interface. Then again, confusingly, the datasheet lists Option 1 as "10x clock multiplier".

[0xdeadbeef]

OK, I was tired of looking for other options, so I bought a Tabor 6010 for 150€ which looks pretty much like in mint condition on the pictures even tough I guess it's most probably from 1992 or so.
The seller confirmed that the TCXO is installed even though it doesn't show option 1 during startup. Can't wait for it to arrive so I can take a look inside

There's another eBay seller who sells a Kontron and (!) a Tabor 6010, both in very good condition but without the 1.3GHz option installed for 123€ (each). As I don't really need the 1.3GHz input, my main reason not to save ~30€ was that he couldn't confirm that the TCXO is installed.

Side note: the 6010 is working in reciprocal mode up to exactly 10MHz (whatever that means). If it really quits reciprocal mode at 10.0000001MHz or so this might be a bit of a nuisance but still I guess that it's quite a bargain for a solid 9 digit counter with two 125MHz inputs, a TCXO and both, GPIB option and the 1.3GHz option installed.

[0xdeadbeef]

So the world is bad. Even though I was explicitly assured by the seller that the TCXO is placed, it simply isn't.
There's just the 5ppm standard oscillator on the PCB which is also connected to the rear panel 10MHz reference output (and input - configurable by jumper).
Besides, using the according trimming capacitor of the 5ppm oscillator changes the measurement so there's absolutely no doubt that the 5ppm standard oscillator is used.
Still the seller claims that I should just open the case and I would see the TCXO in a "black housing". Dunno what he thinks the TCXO would look like but IMHO there is no TCXO and anyway, it's absolutely 100% certain that the 5ppm oscillator is used.
Still, the device in nearly mint condition and works flawlessly. It's hard to believe that it must be at least 20 years old. So I'm hesitant if I'd like to keep it or not.
As a side note, the reciprocal mode doesn't seem to work near 10MHz. At least I didn't get it to work. But I guess I could live with that.

[Ash]

This is a photo of the inside of my 6030, showing the time-base module. You can see that the 5ppm oscillator you highlight is not fit in this case.

Ash.

[0xdeadbeef]

Yeah, this is what I expected to see for the TCXO option
But then again, you don't have the GHz input board

Good thing is that I convinced the seller that he's wrong in the meantime. So I will get a refund this way or the other (either send it back or keep it and get some money back).
I mean it's not totally unusable even with the 5ppm oscillator. I could calibrate it to a few LSDs (i.e. the 9th digit changes driftes counts up and down) and I can't tell if the drift comes from the oscillator or the GPSDO.

While we're at it: did you ever try to jumper the 10MHz output to input? I wonder if it will need the 10MHz reference input all the time then or automatically detects its presence and switches from the internal timebase to the external if there is a signal connected.

[Ash]

Yes, the channel C input is a little bit of a lesson for me 

I assumed it had it because the connector was there, but it wasn't. That in in line with the manual, and the seller didn't say it was fitted when I bought it on eBay. But no matter I got it reasonably cheaply because there was an error with the unit where it wouldn't read correctly. Turned out the PLL that takes the 10MHz and converts to 500MHz (on the 6030, not sure about the 6010) was broken - the first divider was dead. I was able to source a replacement and was all good. I also took the precaution of replacing the electrolytics..

As for the External/Internal switching, I'm not sure about the 6010, but if the 6030 is anything to go my, it is a manual switch - you need to move jumpers around inside the instrument. These are located on the left side of the module.. There is no auto switching  In my case the instrument runs at 500MHz derived from a PLL, so the external 10Mhz would have to be constant..

On my long list of "projects to do when I get a chance" is to break out the jumpers to either a small relay daughter board or even add a toggle switch on the back..  marginal at 10MHz tho.

I also intend to make or re-purpose a divider board to bring the C channel up. I'm reasonably sure there is the ability to go to higher frequencies as some of the pins to the daughter board set the divide ratio so the firmware can correctly display the frequency. the 6030 is labelled at 2.4GHz..

If you have a GPSDO, just setup as an input, and you will be more accurate than me

I don't know what the differences are between the models, but if they share common hardware, I have a copy of the 6030 EPROM if you want to play with it..

Ash.

[0xdeadbeef]

I already have a cheap Chinese GPSDO and used it to calibrate the 5PPM oscillator. However I somewhat dislike the idea of using a GPSDO as (permanent) time base for the counter.
My idea is to create a small PCB with an 10MHZ OCXO plus trimmer, voltage supply and the like and either mount it internally or to the external input (after using the jumper to configure it).
Maybe I could even use the internal jumper J4 to supply the PCB.
Anyway, I understand that if I had the TCXO option (1), the external reference could be 1MHz, 5MHz or 10MHu but without it, only 10MHz is possible (since the PLL 74HCT4046 is on the TCXO board).

One  drawback of this solution compared to the TCXO board is that I won't have the 10x multiplier. With it, the counter is 10 times faster (8 digits per second instead of 7) and the maximum frequency for reciprocal counting is increased to 100MHz. Well, theoretically, I guess I could try to figure out the pinning of J4, create the 100MHZ signal and apply it to the correct pin but I'm not sure if this is worth it.

About the EEPROM: I think the 6030 has higher specs so I doubt  using the 6030 SW in a 6010 would do any good even though the HW looks very similar.

[Ash]

I downloaded the manual for the 6010 (available on the Tabor site), and looked at the pin-out of the connector, its all documented.

If you had a modern 100MHz TCXO (quick digikey search showed some 1ppm versions).. and buffered that you then feed 100MHz into the right pin onto the unit. There are 2 "sense" lines on the header used by the firmware to detect the option and adapt I guess. These are shown in the schematic for the option in the manual.

I'd be betting that it would all work fine, but you would have to disable the existing 10Mhz reference first. That should be as simple as pulling both the int/ext jumpers so there is no connection at all. This will leave the output driver active in one logic state, but this is capacitively coupled to the same line as the 100Mhz input from the TCXO option so hopefully shouldn't be a big issue.

You could probably test by just jumpering the right pins on the header so the firmware detects the option, but leave the 10MHz ref alone, the counter should then be reading 10x high.. If that happens, building your own 100Mhz TCXO should be straight forward. Looks like if you jumper pin 2 (OPT0) to pin 1 (ground) is all you need to do, pin 4 (OPT1) is left floating according to the schematic.

The manual does state that the options are field installable with no software changes needed. And section 5-2 details it.. however it says to pull U58, and the links LK1a/b, but I can't find U58, I'm guessing that they mean U57 which is the 10Mhz reference. Is that socketed?

Anyway - have a look over the manual and schematics etc, and I think you'll see it should be pretty straight forward to upgrade the unit. There is plenty of room inside to mount a TCXO board, or even a OCXO.. however if you have a 10Mhz you'll need to PLL up to 100Mhz. Like I suggested above it may be better to just go straight to a modern 100MHz osc. Guess it depends on how far you want to go. you could do a PLL and have some internal/external reference switching on the board as well.

Ash.

[0xdeadbeef]

Wow, thanks, how could I overlook the schematics in the manual? I actually searched for the Jumper 4 layout in the manual but didn't consider that the schematics were only graphics.
Anyway, I tend to think that the manual mixes up U56 and U58. While U58 is used in the analog output option, U56 is the 74F132 (Quad 2-input NAND Schmitt trigger) used in the 5ppm Oscillator. It's socketed as all the other ICs and I guess it would make perfect sense that the 5ppm oscillator isn't connected to the device anymore if it's removed.

I originally planned to use one of the 10MHz OCXOs I bought some time ago and I guess I could build some PLL circuit to crank it up to 100Mhz.
The benefit of this approach would be that I could also still use an external 10MHz (keeping the 10x multiplier) reference with an internal jumper.
But it's certainly much easier to try to get a 100Mhz TCXO (or OCXO) with <=1ppm. 
Now this really looks like it could be a piece of cake to build myself a fake option 1...

[EDIT]
On a 2nd thought, I guess using some off-the-shelf clock multiplier like the IDT ICS601M-01 I could actually re-use my 10MHz OCXO and keep the possibility to use an external 10MHz Reference.
https://www.idt.com/document/dst/601-01-datasheet
 

[Ash]

Wow, thanks, how could I overlook the schematics in the manual? I actually searched for the Jumper 4 layout in the manual but didn't consider that the schematics were only graphics.

Haha.. I guess I'm used to looking at old manuals which are scans of a photocopy of a printout...

Anyway, I tend to think that the manual mixes up U56 and U58. While U58 is used in the analog output option, U56 is the 74F132 (Quad 2-input NAND Schmitt trigger) used in the 5ppm Oscillator. It's socketed as all the other ICs and I guess it would make perfect sense that the 5ppm oscillator isn't connected to the device anymore if it's removed.

Ah yes, makes sense. I didn't have a lot of time to search for it. Now that you point that out, it would certainly make more sense to pull the NAND that is driving the 10Mhz output. Leave the oscillator there as you may need it at some point and that way you won't loose it

I originally planned to use one of the 10MHz OCXOs I bought some time ago and I guess I could build some PLL circuit to crank it up to 100Mhz.
The benefit of this approach would be that I could also still use an external 10MHz (keeping the 10x multiplier) reference with an internal jumper.
But it's certainly much easier to try to get a 100Mhz TCXO (or OCXO) with <=1ppm. 
Now this really looks like it could be a piece of cake to build myself a fake option 1...

That's the wonderful thing about this older gear - its well documented and can be modified and built upon 

[EDIT]
On a 2nd thought, I guess using some off-the-shelf clock multiplier like the IDT ICS601M-01 I could actually re-use my 10MHz OCXO and keep the possibility to use an external 10MHz Reference.
https://www.idt.com/document/dst/601-01-datasheet

You might want to be careful with that chip - it seems to be designed for direct crystal driving. It also states that the jitter and phase between the reference and output are not specified.

The external input is really useful I think and it is something I'd try to incorporate, but I'd be looking for a higher spec design. The problem is going to be the accuracy of your measurements will be directly affected by the jitter.

Let me know if I can help with anything. I find this lots of fun messing with old gear and I always learn a lot.

Ash.

[0xdeadbeef]

You might want to be careful with that chip - it seems to be designed for direct crystal driving.

It also supports clock in. X2 is just left open then.

It also states that the jitter and phase between the reference and output are not specified.

It's true true that the input to output skew isn't defined but I'm unsure why I would need that for this application.
Anyway, there is also a "zero delay" version available namely the ICS670 which however has a slightly worse phase noise specification (-124 dBc/Hz at 10 kHz for the 670 vs. -132 dBc/Hz at 10 kHz for the 601).

The external input is really useful I think and it is something I'd try to incorporate, but I'd be looking for a higher spec design. The problem is going to be the accuracy of your measurements will be directly affected by the jitter.

The typical output jitter (one sigma) is defined as 12ps for the ICS601 which is probably >100 times less than the time resolution of the counter. Actually I couldn't find any specification for the single shot interval resolution only the 10ps resolution for averaged (!) measurement with option 1 (TCXO). So probably even with the TCXO the single shot resolution is 500ps or more. Actually I think it could be as bad as 2ns.
I don't think a typical OCXO or TCXOs will have much better values regarding phase noise. E.g. I looked up a 100MHz VCTCXO from Abracon (ASGTX-D-100.000MHz-1 for 51,36€) and my understanding is that it has a phase noise specification of -116dBc/Hz at 10KHz for a 120MHz carrier.
So I'm not sure that an affordable TCXO or OCXO will have a much better phase noise or jitter than this multiplier solution.

[Ash]

The external input is really useful I think and it is something I'd try to incorporate, but I'd be looking for a higher spec design. The problem is going to be the accuracy of your measurements will be directly affected by the jitter.

The typical output jitter (one sigma) is defined as 12ps for the ICS601 which is probably >100 times less than the time resolution of the counter. Actually I couldn't find any specification for the single shot interval resolution only the 10ps resolution for averaged (!) measurement with option 1 (TCXO). So probably even with the TCXO the single shot resolution is 500ps or more. Actually I think it could be as bad as 2ns.
I don't think a typical OCXO or TCXOs will have much better values regarding phase noise. E.g. I looked up a 100MHz VCTCXO from Abracon (ASGTX-D-100.000MHz-1 for 51,36€) and my understanding is that it has a phase noise specification of -116dBc/Hz at 10KHz for a 120MHz carrier.
So I'm not sure that an affordable TCXO or OCXO will have a much better phase noise or jitter than this multiplier solution.

You are probably right. I wasn't really thinking through that stuff too much. 

I missed that you could drive the ICS670 with an external clock (I did look, but very quickly). For the price on digikey, I'd be very tempted to just give it a go given you already have a OCXO

Looks like a fun project. I'll certainly be following your progress if you share it!

Ash.

[0xdeadbeef]

Yeah, I'm like 95% percent sure I will do this sooner or later. It's just that I guess I want to design a PCB for this and even if I should do this in the next days it will take >6 weeks or so until I get the PCBs from China. And well, you know if I'm ordering parts it usually ends with days checking what I could order additionally now that I'm ordering anyway.
So don't expect an update on this too soon

[Ash]

Yeah, I'm like 95% percent sure I will do this sooner or later. It's just that I guess I want to design a PCB for this and even if I should do this in the next days it will take >6 weeks or so until I get the PCBs from China. And well, you know if I'm ordering parts it usually ends with days checking what I could order additionally now that I'm ordering anyway.
So don't expect an update on this too soon

We are all busy and sometimes the fun stuff has to wait a bit sadly.. I have a few of those projects myself. 

As for ordering, I'm normally frantically trying to figure out what else I should add to the order to meet the free shipping threshold to Australia.. That often takes much longer than getting the stuff I know I need

Best of luck!
Ash.

[0xdeadbeef]

Yeah, I need to get over 50€ to get free shipping. But I guess this won't be much of a problem - just time consuming

Side note: I looked a bit more into the schematic and it looks like simply connecting a 100MHz oscillator won't do the job.
There seems to be some +/-5V level shifting going on at the output So I guess I will replicate that circuit.

While I'm at it, I guess I will also replicate the 10MHz external reference input/output circuit using the 74F132 NAND-Schmitt-Triggers (funnily they are called Schmidt-Triggers in the manual which is actually the more common German spelling).

[Ash]

Side note: I looked a bit more into the schematic and it looks like simply connecting a 100MHz oscillator won't do the job.
There seems to be some +/-5V level shifting going on at the output So I guess I will replicate that circuit.

While I'm at it, I guess I will also replicate the 10MHz external reference input/output circuit using the 74F132 NAND-Schmitt-Triggers (funnily they are called Schmidt-Triggers in the manual which is actually the more common German spelling).

It appears to be a buffer/driver. On the main board, it is driving 2 ECL based NOR gates. There seems to be a biasing network around U25c(?) which along with the series resistor/capacitor R106/C69(?) for AC coupling is how the 10MHz signal gets through from a standard logic gate..

I can't see Q1 being a level shifter as the output is capacitively coupled (C3). I'm thinking its setup as a buffer and driver from the ECL VCO U7 (MC1648).

Q2 is interesting as well, it seem to be disabling the output if the phase comparator is generating longish pulses meaning that it is not locked. The zener is a 6.2V device, so when the output of U2a NAND goes high it will conduct and allow current into the NPN base and therefore shunt the output signal. If I'm reading that correctly.. pretty clever level shifting happening there

Maybe stick a scope on the line and see what the signal looks like using your current 10MHz osc on the board, then try replicate that. I'm guessing that a modern fast logic gate, capacitively coupled would work. If you have a signal generator you could mock up a driver using a gate and drive it from the sign gen and see the result on a scope.. check the output of the ECL NOR gates as well. Would tell you if it will work.

I don't have any experience with ECL, I'm sure there are others here who could glance at the schematic and tell you for sure what you need

Ash.

[0xdeadbeef]

Well, the output of the normal 5ppm oscillator and the external reference is definitely created vby a TTL NAND gate (74F132). When I follow this signal into the "Measurement Logic circuit", it's simply connected to the ECL NOR gate through a series 820Ohm resistor and 100nF capacitor. So I guess this is exactly what I need to connect my TCXO/OCXO output to the 100MHz input.
Quickly checked this with LTSpice and it looks somewhat promising:

[Ash]

Well, the output of the normal 5ppm oscillator and the external reference is definitely created vby a TTL NAND gate (74F132). When I follow this signal into the "Measurement Logic circuit", it's simply connected to the ECL NOR gate through a series 820Ohm resistor and 100nF capacitor. So I guess this is exactly what I need to connect my TCXO/OCXO output to the 100MHz input.
Quickly checked this with LTSpice and it looks somewhat promising:

Looks like it should work

Have fun!

Ash.

[0xdeadbeef]

Did some schematic design today. Mainly copied the circuitry from the manual and added my OCXO, a 3.3V supply and the ICS601-01 as multiplier.
I also replaced the 74F132 with a MC74ACT132 and the 1N4151 diodes with a (double) BAV99.
I also started with the PCB. It should fit quite easily into a 5x5cm PCB. I'm a bit uncertain if I should solder the shielded wire to the 10MHhz reference input/output directly to the PCB or use some kind of SMB connector.

[Ash]

Cool. Nice work.

The unused NAND U4.4 should have its inputs tied to a rail so they don't float. What about using U4.4 as an additional buffer between the multiplier and the output? If you do that, you may want to add a second decoupling cap to the NAND chip at 1nF or 10nF or both..

As for the input 10Mhz, the TCXO / PLL board in my 6030 has the input coax soldered to pads on the PCB.  This is on the upper left of the board in the photo I shared before. Probably the decider would be how easy it is to add remove the board and if you think you will need to. The SMB is the 500MHz output and the other end is soldered onto the PCB directly (I think it replaces the series resistor from the 10Mhz internal line.

The 100Mhz through the pin header is probably ok, but the 500MHz was clearly pushing it so they didn't use it. Maybe prudent to include pads to launch to a bit of coax and a 0 ohm link to isolate the pin on the connector in case there are signal issues? That way you can jumper the signal to the main board with coax if you need.

Actually, if you just add some accessible ground next to the input of the series cap C1, you could remove that and launch to coax. Then you could use the existing cap on the main board, isolating it from the 10MHz line by pulling the associated series resistor and using its pads for the coax..

Ash.

[0xdeadbeef]

The unused NAND U4.4 should have its inputs tied to a rail so they don't float. What about using U4.4 as an additional buffer between the multiplier and the output? If you do that, you may want to add a second decoupling cap to the NAND chip at 1nF or 10nF or both.

Good point about the floating inputs. Changed that. I'm not 100% sure why I would add another buffer between the multiplier and the output. Or do you mean to be able to output the 100MHz on the 10MHz reference BNC?

As for the input 10Mhz, the TCXO / PLL board in my 6030 has the input coax soldered to pads on the PCB.  This is on the upper left of the board in the photo I shared before. Probably the decider would be how easy it is to add remove the board and if you think you will need to. The SMB is the 500MHz output and the other end is soldered onto the PCB directly (I think it replaces the series resistor from the 10Mhz internal line.
The 100Mhz through the pin header is probably ok, but the 500MHz was clearly pushing it so they didn't use it. Maybe prudent to include pads to launch to a bit of coax and a 0 ohm link to isolate the pin on the connector in case there are signal issues? That way you can jumper the signal to the main board with coax if you need.
Actually, if you just add some accessible ground next to the input of the series cap C1, you could remove that and launch to coax. Then you could use the existing cap on the main board, isolating it from the 10MHz line by pulling the associated series resistor and using its pads for the coax.

Yeah, I saw that the 500MHz is connected directly through coax on your PCB. The PCB on my 6010 seems to look slightly different near the J4 jumper and I kinda trust that the header should work if this is the official approach. But yeah, I guess it makes sense to have a fallback solution in mind. Maybe I could even find the space for another SMA connector that is not populated unless the header connection causes issues.
BTW: is that an SMB connector on your TCXO board? Since there must be like 10 connector types which all look the same (SMA, SMC, SMS and what not) and I'm somewhat uncertain which one to use.
 

[Ash]

What about using U4.4 as an additional buffer between the multiplier and the output?

Good point about the floating inputs. Changed that. I'm not 100% sure why I would add another buffer between the multiplier and the output. Or do you mean to be able to output the 100MHz on the 10MHz reference BNC?

I was thinking between the ICS601 and the output to avoid loading it, but I just checked the datasheet for the ICS601, it has 25mA drive capability so no need. But since you mention it, having another jumper "beside" the output one to send 100MHz to the ref out connector might be useful during testing? How many features can you cram onto the board for fun 

Yeah, I saw that the 500MHz is connected directly through coax on your PCB. The PCB on my 6010 seems to look slightly different near the J4 jumper and I kinda trust that the header should work if this is the official approach. But yeah, I guess it makes sense to have a fallback solution in mind. Maybe I could even find the space for another SMA connector that is not populated unless the header connection causes issues.
BTW: is that an SMB connector on your TCXO board? Since there must be like 10 connector types which all look the same (SMA, SMC, SMS and what not) and I'm somewhat uncertain which one to use.

Fall backs are good But as you say, the "official" ones use the J4 connector so it should be good so I wouldn't worry about a real connector - you wouldn't have one on the other end anyway.

I'm not actually sure what the connector is - I remember it is push on, so not an SMA. I just assumed it was SMB. The parts list in the 6030 manual is helpfully vague: "CON RF MALE JACKSON". It looks very similar to the push on connectors in my HP 8595A spectrum analyser if that is of any help 

As for what connector to use - what ever you have around is best - 100MHz isn't pushing any of them I guess..

[0xdeadbeef]

So that's where I'm standing at the moment. There is no SMB connector in the 3D library, but I will use one anyway I guess.
I actually thought of fancy options like an analog switch (e.g. 74HCT4066) that is controlled by a µC that monitors the external reference input and automatically switches to the external reference if it detects a signal there.
Then again, I have enough unfinished stuff of that kind, so I want to keep it easy and simple.

[Ash]

Nice work. My only suggestion would be to add a screw hole or two for mounting. Make it line up with the slots in the side of the case. My tcxo board is mounted using a screw through the tab of the regulator. You may need to arrange a standoff or something, but the screws just bite into the slot.

Ash.

[0xdeadbeef]

Regarding the hole: it could prove difficult to put it exactly where I could screw it to the case without some trial&error. Probably I'd need to increase the size just to achieve this.
Actually my plan is to either just stick my PCB into J4 and trust in gravity or maybe to 3D print some kind of fixture that I can screw to the case and that just hold the PCB in place from above.

BTW: Reworked it again to use a 74ACT14 (Hex Inverter with Schmitt Trigger) instead of the Quad NAND Schmitt trigger so I can parallel four outputs for the reference output.
Even with four 200Ohm resistors in parallel this is pushing it (especially since these logic gates usually can't deliver more than 100mA in sum - if at all) a bit but the original implementation just uses one output with a 51Ohm resistor.

[0xdeadbeef]

Today it occurred to me that it might be a bit optimistic to be able to draw 800mA or so from the 5V supply (OCXO needs like 700mA during warm-up).
I guess it have to try this before ordering the PCBs. The 5V supply circuit looks somewhat robust though. The bridge rectifier for the 5V tap is rated 5A and the MJE2955A PNP transistor is rated for 10A.
Then again, I have no idea how much is currently drawn from the 5V line. If everything fails, I might actually drop the OCXO idea and use an affordable TCXO instead.

[Ash]

Today it occurred to me that it might be a bit optimistic to be able to draw 800mA or so from the 5V supply (OCXO needs like 700mA during warm-up).
I guess it have to try this before ordering the PCBs. The 5V supply circuit looks somewhat robust though. The bridge rectifier for the 5V tap is rated 5A and the MJE2955A PNP transistor is rated for 10A.
Then again, I have no idea how much is currently drawn from the 5V line. If everything fails, I might actually drop the OCXO idea and use an affordable TCXO instead.

Hum.. good call. I hadn't even through of the OCXO power consumption. The power supply in the 6010 will be pretty solid because the ECL and other "fast" logic they used draws quite a bit of power. I assume it will only be an issue during the initial period of bringing the oven to temperature, then settle down to something lowish.

Could you measure the rail voltage and jumper in the OCXO cold and see if the rail droops? Maybe if you have an adjustable current sink, you could slowly apply the load and see if the power supply will cope?

Ash.

[0xdeadbeef]

My OCXO (C-Mac, LVCMOS, 3.3V supply) draws 700mA for 5 minutes or so, then it drops to ~350mA (at room temperature).
My plan was to use an electronic load to increase the current draw on the 5V line slowly towards 800mA or so and check if the voltage drops.
I'm just too lazy to set up this test today. Chances are this will only happen when I have some more time at the weekend.
Anyway, the bridge rectifier CR28 is quite a beefy part even though it's unclear if it's the 5A (parts list) KBL-005  or the 8A (schematic) version KBU8A.
There are no markings visible on the accessible side but from the looks of it, I would tend to say it must be a KBU type (hole in it), so probably the schematics are correct.
When I'm at it, I will also check the voltages and the 10MHz signal on pin 10 of J4.

[Edit]
What I quickly tested today is the power consumption. It's about 23W in idle ("0" displayed) and ~24W if all LED digits are used. Assuming an optimistic 90% efficiency of the transformer, this would mean like 21.6W on the DC side.
There's a 12V tap, but the +/-12V supply is limited to some opamps and the 5ppm oscillator, so I would assume that the 12V part is probably taking less than 1W ... maybe 0.5W or so. For the sake of convenience, let's say 21W are caused by the +/-5V supply. So by rule of thumb, there could be like ~4A drawn on the +/-5V rail.  This is a bit more than I expected and probably I made a mistake somewhere and/or the transformer efficiency is more like 85% or so.
Still, there probably is a margin for the OCXO, especially if the bridge rectifier is actually the 8A type. If it's the 5A type, the first few minutes would be pushing it a little bit.

[0xdeadbeef]

So I finally got my ass up and did some measurements.

Unfortunately, I wasn't able to get a good measurement of the 10MHz clock at J4. With my 200MHz scope and 2GSa/s sample rate, the signal looked like a rectangular DC signal with a lot of ringing. Then again I didn't dare to use the 50Ohm input but used a 10:1 probe on the 1MOhm input. I guess I'd need an active probe for a proper measurement, but still the voltage levels didn't look nearly like I expected (negative ECL stuff) but more like common CMOS levels.

Secondly, I measured the voltage at the 5V test point while I increased the current draw on the 5V pin of J4 in 100mA steps and also monitored the power consumption at the 230V side.
In a nutshell, the voltage drop on the 5V rail was negligible. Like even at 1A it dropped from 5V to 4.996V. At 300mA the voltage was like 4.998V and dropped to 4.997V at 600mA. No problem there.
At the same time, the power consumption at the 230V AC supply increased by 1W per 100mA. So it increased from 24W at 0mA to 32W at 800mA. Which actually means the efficiency for the 5V rail must be quite exactly 50%!?
Or in other words: the default current at the 5V rail can't be much higher than 2A and there should be more than enough headroom for my 800mA additional current draw.

Finally, I took some thermal images of the bridge rectifier and the transistors of the 5V/-5.2V supply. While the temperature increase on the rectifier looks totally OK (41°C to 44°C), the temperature on the transistor increased from ~48.5°C to 57.6°C. So, yes, there is an increase but it seems acceptable also taking into account that 800mA is about the worst case scenario for the initial OCXO heat up.

As a side note, I also measured my OCXO again. Actually the 700mA/300mA values were from the datasheet and I only remembered the measured values weren't totally off.
So the initial current (at 23.5°C room temperature) was 587mA and after some minutes it settled to ~225mA. 

-> Looks like I'm fine

[Ash]

Unfortunately, I wasn't able to get a good measurement of the 10MHz clock at J4. With my 200MHz scope and 2GSa/s sample rate, the signal looked like a rectangular DC signal with a lot of ringing. Then again I didn't dare to use the 50Ohm input but used a 10:1 probe on the 1MOhm input. I guess I'd need an active probe for a proper measurement, but still the voltage levels didn't look nearly like I expected (negative ECL stuff) but more like common CMOS levels.

If you measured the 10Mhz pin on J4 (pin 10), that is on the logic level side, before the R/C dc block / coupling. I think if you measured the 100MHz "input" pin (pin 6) you would see the ECL voltages as this is on the line that is after the R/C coupling of the 10MHz signal.

If you can use a short spring ground probe rather than the wire lead with clip, you should get a much better picture of the signal. You shouldn't need to use 50 Ohm here - that would cause issues. The signal problems would most likely be related to your grounding connection.

You want to use something like this, putting the ground bit into the nearest ground (next pin on the connector?):

Secondly, I measured the voltage at the 5V test point while I increased the current draw on the 5V pin of J4 in 100mA steps and also monitored the power consumption at the 230V side.
In a nutshell, the voltage drop on the 5V rail was negligible. Like even at 1A it dropped from 5V to 4.996V. At 300mA the voltage was like 4.998V and dropped to 4.997V at 600mA. No problem there.

Good news about the voltage rails for sure!

Finally, I took some thermal images of the bridge rectifier and the transistors of the 5V/-5.2V supply. While the temperature increase on the rectifier looks totally OK (41°C to 44°C), the temperature on the transistor increased from ~48.5°C to 57.6°C. So, yes, there is an increase but it seems acceptable also taking into account that 800mA is about the worst case scenario for the initial OCXO heat up.

Nice. So jealous of the thermal camera - one is definitely on my wish list

Ash.

[0xdeadbeef]

Crap, you're right, I somehow forgot I wanted to measure the "100MHz" pin 6 (where the ECL 10MHz is currently located in my case) and not the 10MHz pin
Anyway, yeah I know the measurement is not perfect since my ground connection sucks but at least the levels look OK now.
About the thermal camera: yeah, it's one of these hacked Flir E4s that I totally wanted to have back then and admittedly haven't used much since then.
So I try to use it now and then just to justify the purchase

[Ash]

Crap, you're right, I somehow forgot I wanted to measure the "100MHz" pin 6 (where the ECL 10MHz is currently located in my case) and not the 10MHz pin
Anyway, yeah I know the measurement is not perfect since my ground connection sucks but at least the levels look OK now.

Hum.. I really hope the signal doesn't look like that in reality.. very slow edges. 
 

About the thermal camera: yeah, it's one of these hacked Flir E4s that I totally wanted to have back then and admittedly haven't used much since then.
So I try to use it now and then just to justify the purchase

haha. Fair enough But surely the justification for test gear is not that you need to use it, but rather that you have it WHEN you need it 

Well, at least that's how I justification my addiction.. um.. I mean investments. 

[0xdeadbeef]

Hum.. I really hope the signal doesn't look like that in reality.. very slow edges. 

My current equipment doesn't allow me to make much better measurements. I still appended two screenshots taken with a 25MHz differential probe.
One shows the 10MHz signal at CMOS level and the other one the ECL level signal. The limited bandwidth sure softens the signal a bit, but I think it's fair to say that the basic shape is still visible.
So yes, the ECL signal has slower edges than the CMOS signal which kinda was to be expected.

haha. Fair enough But surely the justification for test gear is not that you need to use it, but rather that you have it WHEN you need it 
Well, at least that's how I justification my addiction.. um.. I mean investments. 

Yeah, that's exactly what I'm telling myself

[Jay_Diddy_B]

Hi,

Is the ICS601 Obsolete?

Regards,
Jay_Diddy_B

[0xdeadbeef]

What do you mean by obsolete? Like out of production or not needed by my DYI update or like there are much more advanced solutions available?

Regarding the 1st meaning, it's listed "active" on the IDT site (even if the search function there is somewhat dysfunctional):
https://www.idt.com/products/clocks-timing/clock-generation/clocks-general-purpose/601-01-low-phase-noise-clock-multiplier

Regarding my little project, it's not obsolete either since I want to use it for the x10 clock multiplier. Even if I had decided to not use the C-Mac 10MHz OCXO I have in the drawer and bought a 100MHz TCXO, the ICS601-01 would be still be useful to multiply an external 10MHz reference (i.e. GPSDO). Note that the 100MHz internal clock (Option 1) has certain advantages like reciprocal mode working up to 100MHz (instead of 10MHz) and measuering 8 instead of 7 digitis per second.

And should the ICS601-01 be obsolete in the sense there is a better option available, I'd be happy to hear it. Side note: I also checked the 601-02 and 670-01/02 and found the 601-01 seems to have the best phase noise and/or jitter specifications.

[0xdeadbeef]

So nearly a year later I finally finished this little project. I reworked the PCB again, mainly put all the SMD stuff on the bottom to improve the routing.
For the sake of shorter signal paths and easier routing, I also sacrificed the idea to keep both jumpers together.

Still, this DIY upgrade provides the same functionality as the official TCXO update - actually even a bit better due to using an OCXO and an integrated low jitter PLL.
The only thing I haven't done yet is mounting a cable from the SMA connector to the rear panel. Still, the SMA connector can be either used to output the 10MHz OCXO signal or to input an external 10MHz reference that is multiplied to 100MHz as internal clock. Anyway, the TCXO/multiplier option is automatically detected as the board pulls J4/2 low as the original TCXO option. This also means that reciprocal mode is now working up to 100MHz. So full 9 digit resolution now also works for a gate time of 1s (for input frequencies <100MHz).
I also added a mounting hole btw. but don't use it yet. I'd first need to create some spacer with the proper length, but the board sits very tight anyway and couldn't slip out while the case is closed due to its height.

Two remarks:
Unfortunately, I didn't check the orientation when I soldered in the DIP switch. So it's rotated by 180° which means I had to switch position 3 on (connected to GND) to select the 10x multiplier.
Also, just as for the original board, the two jumpers F1a/F1b have to be moved to the "external" position to disconnect the 5ppm oscillator. If this isn't done, it would be connected to the 100MHz output of the TCXO board which would at least totally mess up the internal time base.

[Ash]

Nice work 0xdeadbeef!

Great to see that the project was completed and working.

Cheers,
Ash.

[0xdeadbeef]

Yeah, it took some time. But at least I didn't mess anything major up (apart from that DIP switch being the wrong way around)
Talking of long time projects, as a little side project, I used the 2nd C-MAC OCXO to build a mini 10MHz reference. Actually this is something I began with in 2013 but my first PCB design was kinda screwed up back then and so the idea somehow got lost. But when I did that Tabor OCXO thing, I decided to make another small PCB for that standalone reference and it came out quite nice. Makes me smile when I look at it since it's so tiny and sweet. Not my best work with the file though 

[coromonadalix]

Nice project, congrats   

[Peter1984]

Hello,
inspired by the idea of 0xdeadbeef I made a similar option-1 board for my Kontron 6010 Counter. I use a small 10MHz TCXO (+-0.2ppm) instead of an OCXO but the same ICS601 PLL-IC to generate the 100MHz. It seems to work and the unit shows the new option when booting up. The only problem is that the readings of the counter are now low by factor 10. If I feed 10.0000MHz to its Input the reading is only 1.00000MHz. So it looks like the counter still thinks that it has a 10MHz reference-clock instead of 100MHz.

I can measure the 100MHz PLL output clock at J4-Pin6. J4-Pin2 is shorted to ground and J4-Pin4 is open (measured +4.8V). I removed U56 and the LK1 Jumpers.

Is there any modification I forgot? When going back to the internal 5PPM oscillator the measurements of the counter are fine.

One thing that came to my mind is that my unit only runs on firmware-version 1.2. Is it possible that there is a problem with the old firmware? Are there newer hex-dumps for this model available?

Best regards
Peter

[0xdeadbeef]

Hm. I'm pretty sure I posted everything I did here. My schematics show which pins I pulled low and left open and the picture of the LK1a/b jumpers shows how I placed them.
Besides, the "1.4" displayed should actually mean that options 1 and 4 are installed. Mine shows 1.2.4 as far as I recall. So the software seems to recognize the jumper setting.
Anyway, my software version is 2.1 (label on the EPROM as well as displayed during startup). So I dumped my EPROM in case this could help you.
Note that the EPROM installed in my device is a Dense-PAC D27C256. To read it out, I needed to selected a DPV27C256.
The manual says that "U42 is a 27128 EPROM containing 16K bytes of software". I guess they ran out of 128kbit EPROMs some day and decided to use 256kbit EPROMs with the additional address line pulled high.
So in the 32kByte dump, the lower 16kByte are empty. That why I also created a "27128" version by cutting out the lower (empty) 16kByte. Hope that helps.

[Peter1984]

Wow, that was quick. I will try if my problem is still there with firmware V2.1 and will report on the result.
And yes, you are right. I only have option 1 and 4 (GPIB), so "6010-1.4" is correct. After that it shows "Soft 1.2" for the firmware version.
Thank you for this perfect service.
Peter

[Peter1984]

I upgraded my unit to V2.1 today and the error is gone.  So there is definitely something wrong with firmware V1.2. I renounce to upload the V1.2 firmware since I don't see any reason to use it (unless someone wants to verify the error). The new firmware seems to work without problems on my old unit. And I am quite happy about the performance of the TCXO (Vectron C2260). Of course it is not even close to an OCXO but it is stable within 5 seconds after turning it on.

Thanks again to 0xdeadbeef
Peter

[0xdeadbeef]

Happy to have been useful 

[Hans81]

Hello,

I am aware that the last entry is from 2018.

I would like to rebuild the 1.3GHz option.

Can someone tell me the value of the resistor R24. In the original schematic this value is unfortunately not mentioned.

Thanks a lot

[0xdeadbeef]

Does this help?
BTW: the manual does mention it, but as "SELECTED VALUE"

[Hans81]

This is very helpful. Thanks for the very quick answer.

[EyeDontKnow]

I took a chance and bought a Tabor 6020 parts unit with an "unknown problem".
Before powering it up, I replaced all electrolytics because one was bulging.
With power on, voltages checked good, and it booted as it should (almost, read on...)

 However, with a signal applied at the input(s), error flashes, "No Ref".
It has the factory installed TCXO board.
 TCXO tests good at 10Mhz.
At the ref-board to main-board junction, pin 6 should read 100Mhz.
I'm getting a sine about 115Mhz @ 150mV P-to-P.

For some reason the ref-board is not being recognized by the rest of the system. 
During boot, it only says "6020" , and it should read "6020-1".

[EyeDontKnow]

UPDATE:
 
I believe I got it working.

In the Tabor 6010/6020 manual, "Table 7-4 TCXO and x10 Multiplier Circuit Checks" (tcxo option)... my voltage and hertz readings were way off, especially on pin 6 (output) of the CA3140. It's used as a low-pass filter to the blah-blah-blah. Output on this opamp was  wrong at -12v.... should be about -3.5v.
I looked closely (like really closely with a loupe) and saw some whitish residue under the chip.
I pulled the chip and added a socket. I had 2 old 3140's (used/pulls) so I inserted the cleanest one......but it had the same problem. Darn !!

I also  socketed and replaced  a fresh 74HCT4046 PLL.
Although.... on this factory board they used a 7046 PLL....difference being the 7046 has a "lock detector" output on pin 15 (which on this board is not used (left open).

Still, no change.
Near the end of my rope, I switched that CA3140 to my last and ugliest spare chip.
Bingo..... no more -12v at the output......and a good 100Mhz @80mV sent out of the option board. It now measures freq, and without the "No Ref" error.
The counter now works, though the boot-up display still does not indicate the option "1".
I'll have to do a thorough Performance Check.

=R=