184 marker generator problems


Morris Odell
 

Hi all,

I'm trying to repair a 184 that appears to have suffered at the hands of a previous owner. The delay line has no ferrite beads and is not working but that's not such a big deal as it can be bridged to allow markers to get through even if they are not coincidental.

The immediate problem is in the 10 MHz to 2 MHz (0.1 uS to 0.5 uS) divider. This is a charge pump circuit where each cycle of the 10 MHz reference input pumps a little bit of charge out of a fixed 15 pf ceramic cap into a storage cap (39 pf in parallel with a 9-50 pf ceramic trimmer) and when it gets to a certain voltage after 5 pulses it triggers a blocking oscillator using a 2N964 high frequency Ge pnp to provide 2 MHz pulses. In my unit it divides by a larger number than 5 even at the extreme of the trimmer cap setting. That indicates either that the storage cap has gone high or the 15 pf cap is too low in value. Alternatively one of the two isolating diodes might have too much reverse leakage which is probably more likely than a ceramic cap drifting. The 2N964 tests OK and temporarily replacing it with a high frequency Si transistor doesn't make much difference. Either way it will be difficult to replace the appropriate bits. That divider affects all the other marker frequencies which are unstable and incorrect.

Before I unplug lots of wires to the board, replace diodes and disconnect the delay line I wonder whether anyone else has encountered this problem or has any advice about it. There's no rush as I'll be busy with other stuff for a couple of days anyway.

Thanks,

Morris


Tom Gardner
 

On 04/08/21 08:15, Morris Odell wrote:
Hi all,

I'm trying to repair a 184 that appears to have suffered at the hands of a previous owner. The delay line has no ferrite beads and is not working but that's not such a big deal as it can be bridged to allow markers to get through even if they are not coincidental.

The immediate problem is in the 10 MHz to 2 MHz (0.1 uS to 0.5 uS) divider. This is a charge pump circuit where each cycle of the 10 MHz reference input pumps a little bit of charge out of a fixed 15 pf ceramic cap into a storage cap (39 pf in parallel with a 9-50 pf ceramic trimmer) and when it gets to a certain voltage after 5 pulses it triggers a blocking oscillator using a 2N964 high frequency Ge pnp to provide 2 MHz pulses. In my unit it divides by a larger number than 5 even at the extreme of the trimmer cap setting. That indicates either that the storage cap has gone high or the 15 pf cap is too low in value. Alternatively one of the two isolating diodes might have too much reverse leakage which is probably more likely than a ceramic cap drifting. The 2N964 tests OK and temporarily replacing it with a high frequency Si transistor doesn't make much difference. Either way it will be difficult to replace the appropriate bits. That divider affects all the other marker frequencies which are unstable and incorrect.

Before I unplug lots of wires to the board, replace diodes and disconnect the delay line I wonder whether anyone else has encountered this problem or has any advice about it. There's no rush as I'll be busy with other stuff for a couple of days anyway.
It might also be that there is too little charge being put onto the 15pF capacitor, perhaps due to a (carbon composition) resistor being too high.

Tacking a resistor in parallel with a suspect resistor could be a way of testing that without removing the board.


Morris Odell
 

After reading Tom's post I did look at the waveforms in the divider circuit to check that it was transferring the right amount of charge. Sure enough the levels were low all the way through. The divider gets its 10 MHz signal from a link winding over the main oscillator tank coil. On close inspection it looked like someone had been fiddling with the number of turns on the link. There were 3 turns on it and when I tacked in another turn everything came good! I did have to bypass one side of the delay line and replace one dud transistor down the divider chain but it's working well.

Now the question is, what is the right number of turns for that link? If it's 3 then maybe the oscillator tube is weak. If it's 4 then no more work needs to be done. I would be very grateful if someone with a working 184 could take a look and let me know. The osc tube is a 7587 tetrode nuvistor - not exactly easy or cheap to replace!

Thanks,

Morris


Tom Gardner
 

Mine has 2.5 turns, i.e. 3 visible on one side, 2 on the other.

The whole design is rather fragile. I had to bend the loop of L69 so it was much closer to the stirrup of L70.

I did buy a 7587 from this person https://www.ebay.co.uk/itm/310087990630 He sent the wrong one, but sent the correct one quickly and wasn't interested in having the incorrect one returned.

On 05/08/21 00:28, Morris Odell wrote:
After reading Tom's post I did look at the waveforms in the divider circuit to check that it was transferring the right amount of charge. Sure enough the levels were low all the way through. The divider gets its 10 MHz signal from a link winding over the main oscillator tank coil. On close inspection it looked like someone had been fiddling with the number of turns on the link. There were 3 turns on it and when I tacked in another turn everything came good! I did have to bypass one side of the delay line and replace one dud transistor down the divider chain but it's working well.

Now the question is, what is the right number of turns for that link? If it's 3 then maybe the oscillator tube is weak. If it's 4 then no more work needs to be done. I would be very grateful if someone with a working 184 could take a look and let me know. The osc tube is a 7587 tetrode nuvistor - not exactly easy or cheap to replace!

Thanks,

Morris


Christopher Hilton-Johnson
 

Morris

I am in UK. & 'should be' travelling to US in November.

If you want me to get you a couple of the 7587 nuvistors, from the UK based ebay seller, let me know & I will organise it

Chris HJ

On 05/08/2021 00:28, Morris Odell wrote:
After reading Tom's post I did look at the waveforms in the divider circuit to check that it was transferring the right amount of charge. Sure enough the levels were low all the way through. The divider gets its 10 MHz signal from a link winding over the main oscillator tank coil. On close inspection it looked like someone had been fiddling with the number of turns on the link. There were 3 turns on it and when I tacked in another turn everything came good! I did have to bypass one side of the delay line and replace one dud transistor down the divider chain but it's working well.

Now the question is, what is the right number of turns for that link? If it's 3 then maybe the oscillator tube is weak. If it's 4 then no more work needs to be done. I would be very grateful if someone with a working 184 could take a look and let me know. The osc tube is a 7587 tetrode nuvistor - not exactly easy or cheap to replace!

Thanks,

Morris




Morris Odell
 

Thanks Chris,

If you want me to get you a couple of the 7587 nuvistors, from the UK based ebay seller, let me know & I will organise it
I'm locked down here in Australia and shipping from most places and the USA in particular is extremely expensive. However I have bought a couple of cheap Russian tetrode nuvistor equivalents which should do the job if necessary. They have pretty much the same numbers in the datasheet but have solder leads rather than a nuvistor base. I'm hoping they can be plugged into the nuvistor sockets like the Tek transistor sockets.

At the moment the 184 is working perfectly so I'm reluctant to fiddle with it further.

Cheers,

Morris


Albert Otten
 

Mine has 2.5 turns, i.e. 3 visible on one side, 2 on the other.
Mine the same Tom. Colored blue/white.
Morris, if a previous owner fiddled with this, then probably also with the cores of L18 and L19 (and who knows L15/16 as well). You might try to recalibrate these.
Albert


Christopher Hilton-Johnson
 

Morris

OK not a problem - age and general mental decrepitude made me think you were in USA!

Chris

On 05/08/2021 10:49, Morris Odell wrote:
Thanks Chris,

If you want me to get you a couple of the 7587 nuvistors, from the UK based ebay seller, let me know & I will organise it
I'm locked down here in Australia and shipping from most places and the USA in particular is extremely expensive. However I have bought a couple of cheap Russian tetrode nuvistor equivalents which should do the job if necessary. They have pretty much the same numbers in the datasheet but have solder leads rather than a nuvistor base. I'm hoping they can be plugged into the nuvistor sockets like the Tek transistor sockets.

At the moment the 184 is working perfectly so I'm reluctant to fiddle with it further.

Cheers,

Morris






Morris Odell
 

My thanks to all those who replied about this. The Russian 6 3 12 H-B (english transliteration 6Z12N-V) equivalent tubes arrived yesterday. I got 8 of them including shipping for the cost of one genuine 7587 excluding shipping. They look the same as genuine RCA nuvistors and the electrical characteristics look very similar but they have solder-in leads rather than pins and lack the asymmetrical locating lugs on the bases of real nuvistors. The base pin/lead arrangement is the same. Date codes are 1979 and 1983. The leads looked like they had been soldered on to base pins but they turned out to have been spot welded. I had to carefully clip them off above the welds so they would fit into a nuvistor socket. Once I did that and replaced the oscillator tube it worked just the same as with the original 7587. However despite tweaking the various coli slugs and trimmers it still needs the extra turn on the link to work properly. The oscillator frequency once the oven has warmed up is within 10 Hz of 10 MHz (1 part per million) which is more precision than I'll ever need.

I'm going to leave it like that. It's still a perfectly usable and beautiful instrument. I don't need a marker generator often but when I do, I now have a choice between the 184 and the later 2901.

Morris


Eric
 

My 184 will also drift up and down. When doing the calibration I have the
set point centered at 10 mhz. But the crystal will drift up and down about
15 hz due to the xtel oven heating and cooling. This is normal acording to
the spec sheet and within spec so totally happy with it. This effect is
observed hooking a frequancy count to the crystal out. Kind of cool. But
really hard to adjust with it constently moving.

Eric

On Wed, Aug 18, 2021, 1:44 AM Morris Odell <vilgotch1@gmail.com> wrote:

My thanks to all those who replied about this. The Russian 6 3 12 H-B
(english transliteration 6Z12N-V) equivalent tubes arrived yesterday. I got
8 of them including shipping for the cost of one genuine 7587 excluding
shipping. They look the same as genuine RCA nuvistors and the electrical
characteristics look very similar but they have solder-in leads rather than
pins and lack the asymmetrical locating lugs on the bases of real
nuvistors. The base pin/lead arrangement is the same. Date codes are 1979
and 1983. The leads looked like they had been soldered on to base pins but
they turned out to have been spot welded. I had to carefully clip them off
above the welds so they would fit into a nuvistor socket. Once I did that
and replaced the oscillator tube it worked just the same as with the
original 7587. However despite tweaking the various coli slugs and trimmers
it still needs the extra turn on the link to work properly. The oscillator
frequency once the oven has warmed up is within 10 Hz of 10 MHz (1 part per
million) which is more precision than I'll ever need.

I'm going to leave it like that. It's still a perfectly usable and
beautiful instrument. I don't need a marker generator often but when I do,
I now have a choice between the 184 and the later 2901.

Morris