Transmitter Drive Unit Type MA79H

This is the third MA79H that I have restored to working order. This one differs from the others in that it incorporates an LA255 Tone-to-DC Converter and the additional three fuses in the distributed HT lines. Unlike the 'Fully Loaded' MA79G that I previously put together, the MA79H does not provide a 1.4MHz output or provision for an external 1.4MHz input. The MA79H is specifically intended to be used with the MA350B Decade Frequency Generator, and thus includes an MA284 Mixer in place of the 1MHz crystal and associated oven. Like the RA17 series of receivers, the MA79 series of drive units employ the Wadley drift-cancelling system. The inclusion of the MA284 forms a second Wadley drift-cancelling system resulting in a super-stable signal source.
Aesthetically, this particular MA79H is different from others in that it is fitted with the Perspex finger-guards normally only found on RA17 MKIIs. I suspect these were fitted after it's time in service.




Right: Although the manual refers to a dust-cover, I've never come across an MA79 where the rear is fully enclosed. It is likely that if there were such a thing, it would be automatically removed prior to installation in a rack system ... The MA79 generates a massive amount of heat. In this one, the small perforated cover at the front was slightly deformed and the warning label on top of the fuse-holder sub-assembly was missing. Also, this is the first time that I have encountered a piece of Racal kit with a full complement of Allen keys and trim tools!







Like the RA17 etc., the two VFO modules must be removed in order to access the enclosed circuitry. What is referred to as the Modulator Assembly, but is in fact a whole lot more, comes away much like the 100KHz I.F. module in the RA17. The complex Calibrator Assembly can also be removed ... easier than I expected, and makes working on it so much easier. With the MA79 stripped down to this level, it is then that you notice the actual level of grime that accumulates. I also used this time to check out all the Pye Pattern-12 connectors on the the coax leads. Coax was not widely used until the late 1940s and while North American manufacturers attempted to standardise by coming up with the PL-259 and SO-239 and B.N.C., in the UK it was left to the individual manufactures to come up with their own, often infuriating coaxial contraptions ... hence Burndept connectors and the Pye Pattern-12. I found several of the latter where it was questionable if there was an adequate contact with the outer braid. These were subsequently re-made.






The LA255 and Fuse Holder assemblies are stacked on what Racal called the rear panel. While the chassis is stripped down, now is the best time to remove these two assemblies in order to gain access to the grime accumulated underneath on the main chassis. I also cleaned all the coax cables and tested the valves on the main chassis at this point. I found that there was an alarming low resistance between the Heater and Grid 1 of V1 (EF91), which was duly replaced.

I then decided to tackle the circuitry associated with the mains input and the distributed power supply. Unlike the RA17 etc., the PSU in the MA79 is not confined to a single compartment.







Above: Before and after photographs of the Mains Input Filtering, which with the inclusion of two chokes and two 100nF capacitors, is significantly different to that in the RA17 etc. The components around the Burndept connector on the right are the R.F. detector for the power meter. The panel on which all these components are mounted is thankfully removeable.



Above: On the left; C199/10nF (centre), across the AC input to the bridge rectifier (D4, D5, D6 and D7). It is not uncommon for the original Hunts capacitor here to either blown out one end, or more spectacularly to simply rupture in the middle. Here it is replaced with a modern polypropylene type rated at 630V.

On the right; The additional 7W 27K wire-wound resistor (designation unknown) is wired across the bulk 250V HT line and most likely serves to load the HT line at switch-on prior to all the valve heaters reaching full temperature, thus providing some protection against possible cathode stripping.



Above: Before and after photographs of the MHz VFO and PA circuits. From this angle, this module might be initially mistaken for the 1st VFO in a MKI RA17. The compartment on the left enclosing the VFO (V23) and the mixer stage (V20) is almost (but not quite) identical to that in the RA17. The compartment on the right houses the amplification stages (V26, E180F and V30, EL821). Apart from the layout of the tuned inductors, the circuitry here bears no resemblance to the RA17 VFO module.
This is an easy unit to work on and I didn't encounter any issues other than when I tested the valves on my VCM163, V23 (VFO, EF91) and V30 (PA, EL821) both failed. The EL821 had a tendency to exhibit anode-current runaway!
After replacing the Hunts capacitors and all the carbon resistors (1/4W and 1/2W) with 1W types, the module, which fits my RA17 1st VFO test jig, was thoroughly tested and found to be within spec. I deliberately did not make any adjustments to any of the ten tuned inductors (five pairs). Unlike the alignment procedure for the RA17, Racal did not include specific alignment instructions in the user-manual for this module. My plan was this; If once the MA79H was fully tested and within spec. across the range of 1.5MHz to 30MHz, I would set the Output Tuning control to 5 and see at what frequency per Output-Range the R.F. output peaked. In theory, this should tell me the frequencies at which to adjust the pairs of tuned inductors.



Above: Before and after photographs of the KHz VFO. This is where things became interesting. Like the MHz/PA module, this is not a difficult unit to work on, and I quickly had all the Hunts capacitors and carbon resistors replaced with modern types. I have a genuine Racal test-jig for these things ... fits all KHz VFOs, but it's just as easy to test it lying on the bench. And that's when I noticed something wasn't right. The VFO side of things was working perfectly, but the crystal oscillator side simply didn't function. All MA79s come with three choices of KHz VFO source. These being; External, VFO and Crystal, and of the latter there is provision for six preset crystal controlled frequencies. I have yet to encounter an MA79 KHz VFO module with any crystals in it, and I only have one HC6U crystal which is just outside the frequency range. No matter what position the crystal was in, there was no output from the 6AS6, yet the voltages around the valve were all in spec. Then whilst checking the continuity between the crystal sockets and the 6AS6, I noticed that the wiper of SA1F was not connected to the cathode of the valve. In the above photographs, this is the white insulated wire that runs between the switch wafer and the tag strip. This is a fairly solid tinned copper wire with a silicone-type heat-resistant insulation over the top. Whereas this material is highly resistant to temperature, it has a tendency to tear rather easily if pulled. And that's what I thought had happened ... see the arrow in the photograph above. I had noticed the tear in the insulation whilst working on the unit and as such a tear is a common sight, I had ignored it. However, 'twas not a simple tear. Closer examination revealed that someone had deliberately disabled the 6AS6 crystal-oscillator by cutting this wire. Once the wire was reconnected, the crystal oscillator functioned perfectly.


Whilst refitting the VFO modules to the chassis, I investigated a wiring anomaly that I had noted when taking the modules off the chassis. See the photograph on the left. This is the relay mounted on the rear of the film-strip carrier, which delivers the 110V supply to Oven 1. Note the cables coming into the frame top and bottom, and how in each case, the red wire is folded back and isolated. The one at the top is more noticeable and carries the actuating voltage to the relay, whilst the cable coming in bottom left carries the 110V for Oven 1 ... the KHz VFO oven. Not only had the crystal oscillator been disabled, but so was the oven associated with the entire VFO module. There is of course a valid reason for this ... this being after all, an MA79H designed specifically to be 'married' to an MA350B, which as well as supplying the 10MHz and 200KHz for the MA284 and the 1MHz, it also provides a synthesised signal between 3.6MHz and 4.6MHz to replace the internal KHz VFO.

I have an MA350B which belongs to the owner of this particular MA79H, but I have yet to verify its integrity, so until then I will use my 'Poor Man's' MA350. And for the whole thing to function, it would be wise to have a fully function internally generated VFO with temperature regulated oven. Thus, I re-enabled the relay and reconnected the 110V ... and everything worked!



Above: Removing the Calibrator assembly from the chassis is easy and makes is so much easier to work on. A detailed description of the Calibrator can be found in the first article in this series (see top of page). Suffice to say that this assembly is a critical component in the MA79, which requires the 100KHz divider to run at all times in order to generate the 1.4MHz primary source, upon which everything is based.

Right: The tag-board with the large yellow capacitor is part of the final stage in the 1.6MHz calibrator.

Speaking of the Calibrator ...

I generally do not have any problems with the regenerative divider method used to generate the 100KHz and 10KHz signals. However, although it is relatively easy to configure, the 100KHz divider can prove a tad fickle. Some people have even been known to simply remove it from their RA17. As long as you follow the instructions in the RA17 manual, setting up the 100KHz divider is easy. Of course, this requires you to have a suitable signal source and an oscilloscope. I have to date, now aligned close to seventy of these circuits, and one thing that is a constant is this: Setting it up on the bench is only half the process. I guarantee that you will always have to tweak it once it is installed. After that, switch it off, then back on again and check that it starts. The reason for this is because sometimes tuning each of the cores for a maximum signal can result in too high a Q, such that the regenerative side of things fails to 'fire up'. At this point all you need to do is back off (a quarter turn or so) L45 in the MA79 (L75 in the RA17 etc.). Nine times out of ten, you can consider it done, and it will run for years without needing re-adjusting. However, not all calibrators are equal. Quite often I have encountered one which although it works, the tone that it generates is weak and 'insipid'. This might prompt some people to reach for the trim tool and 'tweak away', and inevitably upset the nature of the circuit. Exactly what causes a minority of calibrators to do this is a mystery to me ... but I am prepared to surmise that it might be down to the valves themselves.


Above: Before and after photographs of the inside of the MA284 Mixer Module. Since this is a small diecast box, space is a premium, and is probably why Racal chose to use Mullard polyester capacitors for C5 and C9. These are the striped ones that look like sweeties and are notorious for their legs falling off. There is nothing in the way of an alignment procedure for this module. However its function is relatively simple. L1 is tuned to 10MHz while L2 is tuned to 10.2MHz. The purpose of the MA284 Mixer is to dramatically improve the frequency stability of the Drive Unit (in this case, the MA79H). This is achieved by eliminating one of the crystal oscillators. V5 (EF91) which is normally configured to produce a 10MHz signal is disabled. The output from the MA79H's 10.2MHz oscillator is mixed with the super-stable 200KHz signal from the MA350B to generate a more stable 10MHz signal. The anode of V2 (EF91) in the MA284 is connected to the anode of V5 where L13 is tuned to 10MHz. L2 in the MA284 serves to attenuate any signal at 10.2MHz.
A switch on the MA284 allows the mixer to be disabled. In this case, the switch can be set to 'OUT'. This reconfigures V5 in the MA79H as an oscillator. However the MA350B is still required to deliver the 1MHz signal as before. This feature is required when the MA79H is in FSK mode.


Above: Before and after photographs of the LA255 Tone-to-DC Converter module. No information is available on this unit. However I believe it to facilitate a form of VOX, or automatic keying. The input of this unit is connected directly across the MA79's A.F. Gain control. The output is connected to the junction of D2 and D3 and thus to the control grid of the C.W. Keyer (V12). Application of an audio signal to the input of the LA255 does generate a DC voltage. However I have not been able to find a way to generate a voltage which facilitates keying.


Above: Before and after photographs of the Modulator assembly. These show the assembly in place, on the chassis. As previously said, the Modulator is attached to the main chassis much like the 100KHz I.F. in the RA17 etc. and as such was removed while being worked on. There are 12 valves in total on this assembly. Only two and a half of these, V9 (EF91), V13 (EB91) and V6b (12AT7) can be described as the Modulator. V19 and V5 are oscillators, V2, known as the Reactance Valve is used to apply FSK and V16 is a curious little circuit known as the Keying source. All the rest form an elaborate ALC and I.F. chain which takes the base 1.4MHz signal from the Modulator and delivers a signal between 2MHz and 3MHz to the final mixer in the PA/MHz VFO module. It was whilst testing the valves on this assembly that I discovered that V22 (6BA6), the ALC valve and V25 (6BE6), the 11.6MHz mixer were in fact swapped. I also noticed that Oven 3 was not the specified Cathodeon type but had been changed to one manufactured by Snelgrove. This is not the first instance that I have found Oven 3 to be a different model. Whereas the Cathodeon oven is a dual voltage type, allowing it to be run from 12V or 6V, the Snelgrove oven is only internally wired for 6V and the Octal socket had been re-wired accordingly.
Because of the compact nature of its construction and the fact that there are three variable capacitors and a wafer-switch nearby, the Modulator assembly is perhaps the most difficult assembly in the MA79 in which to change components.


Above: Still on the Modulator ... Due to the lack of space around the valve bases, much of the circuitry is distributed around five tag-boards of various sizes. This is the largest of these. The axial capacitor on the left is C99 (12uF/250V). Since 12uF is not a 'preferred' value these days, I have in the past replaced this with a rather expensive axial 15uF/450V capacitor by Rifa. However I noticed that the component layout diagram Fig. 10 shows C99 as a radial capacitor ... and I had a bag of radial 15uF/450V capacitors from a previous job! All that was required was a short red wire and the considerably smaller radial 15uF capacitor fitted snugly on the end of the tag-board.


Above: Just for fun ... three photographs showing the progressive re-build of the MA79H.

At this point in the refurbishment process I felt that it might be worthwhile performing a controlled power-up. At this stage the under-chassis compartments had not been touched, and the Modulator was yet to be tested. I refitted the front panel and connected my patching panel to PL5 with pins 3 & 4 and 5 & 6 linked as per the manual. Then after setting the MHz VFO to 3 and the KHz VFO to 500, setting the Transmission Selector switch to CW, the Input selector switch to OPERATE and the Calibrate selector switch to OFF, I verified that there were no unusually low resistance across the various HT lines then powered the MA79 up via a variac, using my Poor Man's MA350 to supply the 1MHz and 200KHz signals. Not that I expected any, but there was no smoke, which is always a good sign.

With the Meter Selector set to R.F. Level and the Output Range switch set to 3 - 6, I turned the R.F. Gain switch fully clockwise and whilst adjusting the Output Tuning control and observing my power meter and the front panel meter on the MA97, I 'hunted' for evidence of power being developed ... which there was ... another good sign. Not only that, after carefully adjusting the Output Tuning and the MHz VFO, the power meter was showing in excess of 500mW and monitoring the output on an oscilloscope, the signal was clean. The frequency was about 50KHz out across the tuning scale but that was easily corrected by adjusting the position of the film-strip in relation to the main drive sprocket.

I turned the R.F. Gain control back and confirmed that when the panel meter was at 0dB, the actual output power was nominally 100mW. However ... when I set the MA79 to generate Double Sideband, there was no modulated signal ... something was not right with the modulator. Having experienced modulator issues with a previous MA79 where the associated tag-board had been assembled upside down, I was half expecting the same fault. But that was not the case this time, as all the voltages around the 12AT7 were correct.

See the photograph on the right. The fault turned out to be a failed solder joint. When running the 12AT7 heater from 6.3V, pins 4 & 5 are tied together and the centre point grounded. In this case, the heater supply was wired to pin 4 with a link to pin 5. However pin 4 had not been correctly soldered. The consequence was rather ironic in that the half of the valve that was NOT used was the only half that was heating up. I cleaned and resoldered the joint at pin 4 and the problem was resolved.



With the used side of V6 now heating, the modulator was easily set up to generate Double Sideband. Although my audio signal generator cannot deliver the level specified in the manual for testing the modulator, I was able to achieve a carrier null of 35dB such that when Upper or Lower Sideband was selected, there was no evidence of residual carrier. I then worked my way through the alignment procedure in the manual ... all performing well, until I came to the section where the switch on the MA284 is set to OUT. When FSK is the selected mode the MA79H's internal 10MHz crystal oscillator is enabled. I found that there was no 10MHz signal on pin 1 of V8, the 1.6MHz mixer.


As previously mentioned, Oven 3 was not the one specified in the parts list and the Octal socket had been re-wired to suit a dual-HC6U Snelgrove type. I then noticed that there was a black wire from pin 5 on the oven to the adjacent insulated terminal, which was in turn connected (the orange wire) to pin 7 (screen grid) of V5. Pin 5 on the Snelgrove oven appears to be internally connected to pin 7 which is wired to ground. Thus V5 was effectively disabled. I suspect that since much effort had gone into disabling elements of the KHz VFO that the internal 10MHz oscillator was likewise deliberately disabled. Maybe this particular MA79H was not intended to generate an FSK signal. Or maybe someone was simply having a bad day.

However, transferring the black wire from the insulated terminal over to the earth terminal adjacent to V5 did not result in a functioning oscillator. The voltages around the valve were 'nominal' as NASA would say. The physical location of this valve is a tad awkward, being under the edge of the adjacent side-panel whilst the underside of the valve socket is very close to switch-related metalwork. As far as I was aware, the valve was heating up but where it was situated, I could not see a glow. I removed the valve and it tested excellent on the VCM163. However, when on the tester, when the heater was activated it exhibited the very bright surge common to a lot of valves. When in the MA79, it did not do this, yet it was getting warm. In actual fact, this was looking like the second bad solder joint in the heater chain is this particular modulator. But it wasn't. Cleaning and resoldering the joints on pins 3 and 4 did not make any difference. Occasionally wiggling the valve resulted in the heater only occasionally lighting and applying de-oxit to the socket made no difference. So I bit the bullet and replaced the socket ... not an easy task by any stretch ... but better to replace something that was intermittent. This did the trick, obviously, and I was able to complete the alignment without any further issues arising. In hind-sight, the valve was getting warm from its proximity to the oven and V19.









Having completed the alignment procedure, I cleaned all the knobs and re-whited them where necessary. This is the first MA79 that I have seen where the dial-brake and Xtal-VFO switch knobs have a white ring on them. I know how shallow the groove is so I was very careful when cleaning them not to dislodge any of the white filler. If the 2MHz to 3MHz and/or the 40MHz filters are not wide enough, the output tends to drop off below spec. at either end of the KHz scale. However, I was very happy with the output of the MA79, so I had no worries about the filters. I had verified the 2MHz - 3MHz filter when I had the Modulator Assy. on the bench. I used my Rigol spectrum analyser and tracking generator to check the shape and response of the 40MHz filter. There was about 6dB of ripple across the pass-band which I managed to get down to 4dB and I also managed to reduce the over-all insertion loss ... not that it made any difference to the output power of the MA79.

The small label, bottom left on the front panel is marked FSK/CW Condition 1 & 2, and covers over the original annotations which were Mark & Space. I don't think this is indicative of a modification, more likely it is a clarification. Essentially Mark and Space when in FSK mode translate to Key UP and Key DOWN in CW mode.

Above: Just for show ... two self-explanatory photographs of MA79H Ser. 559. I find it amusing that such a complex piece of equipment is only intended to deliver 100mW of R.F. When this one was running on the bench, using my Poor Man's MA350 to provide the 1MHz and 200KHz signals, there were no less than 41 valves running. But then you have to take into account that this MA79H was manufactured in January of 1967 ... it can produce a rock-steady signal of CW, AM (DSB), SSB (USB or LSB) or FSK, all at a time when SSB was a 'New' mode. I think Racal did a very good job with the MA79.

Finally: As previously mentioned, Racal did not, for some reason, provide the end-user with alignment instructions for the Power Amplifier stage. Whereas they provided clear instructions on how to align the preselector in the front-end of the RA17 and all its derivatives, we get nothing for the PA in the MA79. The PA stage has five pairs of tuned transformers where the input and output are tuned on each of the five ranges. It is vitally important that each of the transformer pairs is tuned to the same frequency, otherwise there will be a double-hump response.

It was always my intention to come up with a list of initial settings for the PA, and since I did not realign the PA in this MA79H, and the output proved to be well within spec. over the entire range, I have concluded that the following table provides a very reasonable set of initial settings if you have to realign the PA.