Electronic mechanical vibrator substitutes

[Argus25]

Hello,

I thought I would start this thread, not only to draw attention to one of the greatest military radios ever made (The New Zealand made ZC1 MK2) but to a range of possibilities for electronic replacements for the conventional electro-mechanical vibrator.

This article contains 4 different variants with efficiency figures:

http://worldphaco.com/uploads/ZC1_MK...R_SUPPLIES.pdf

Some of the commercial units that do this contain Cmos gates (IC's) wired as square wave oscillators and Darlington or Mosfet output transistors.

One type of unit described in the article contains nothing more than a pair of germanium power transistors and a feedback transformer.

One thing to be aware of is the "leakage reactance problem" outlined in the article, so even with some electronic units, an additional tuning capacitor is required.

In any case I hope this article is of help to those with transceivers which require an electronic replacement.

Best regards, from Australia.

[G6Tanuki]

"Vairy Interestink!" as they used to say, seeing how different people have approached the vibrator-replacement issue.

I've got a couple of 1950s "R209" HF receivers whose vibrators are a bit problematic - so over the winter I'll be investigating the Australian Solution!

Traditionally, one of the big problems with transistor inverters using a simple feedback winding was that the feedback reduced with increasing load: with little load you had lots of feedback just when you didn't need it but at full load the feedback fell to the point where the transistors were not necessarily switching 'hard' so the efficiency fell away. Sometimes a heavily-loaded inverter wouldn't start: I've seen various 'solutions' to this involving slugged relays.

[Station X]

This thread describes the solid state vibrator used in my own ZC1:-

https://www.vintage-radio.net/forum/...ad.php?t=41944

[Argus25]

Station X,

Yes it is interesting to look at all the possible variations on the theme. In the case of the circuit that simply uses 200 or 220R feedback resistors this type sometimes tends to be unstable at RF frequencies depending on the transformer and requires 0.1uF base-collector miller capacitors to stabilize it when 2N3055's are used. There was a version of this published in Electronics Australia in the mid 1970's, that used 470uF coupling capacitors in series with the feedback resistors and 1n4001 diodes backwards across the B-E connections of the 2N3055's to prevent them zenering. These versions have high dissipation in the feedback resistors and its better to use a feedback transformer.

One advantage of Germaniums is that they are very stable due to their poor performance at high frequencies and they have a low C-E saturation voltage drop. On the other hand its hard to beat a low Rds on mosfet and they have very low gate drive power requirements.

In general after years of experimenting with different types, in the case of an oscillator driven output stage the leakage reactance issue sited in the article is more problematic and additional tuning capacity is required. For self oscillating units, the overshoot at switching is much lower. One other advantage of the self oscillating units is that they makes the supply intrinsically short circuit protected, but as noted by G6Tanuki if the supply is heavily loaded to start it might not start as the feedback is reduced. However in tube gear this is seldom the case due to tube warm up times and low initial HT currents.

Two other variants not published in the article include one with a UJT oscillator and another with a unique circuit with an oscillating 7474 flip flop (its an interesting trick to turn a TTL flip flop into an oscillator).

I also tried a unit using a switch-mode psu IC and another logic gate variant where the contact open/close time of the mechanical unit was replicated, but nothing was gained from this except a small drop in duty cycle and reduced output voltage.

One other issue noted in the article is it pays to have much higher PIV rated diodes (if one is replicating a synchronous unit) than calculation suggests. This is because if there is a bad socket connection or the unit is unplugged while running, the collapsing field in the transformer can generate very high voltages and cause rectifier failure.

It is good to hear of ZC1's alive and well in other places in the world.

[G6Tanuki]

The issue I once had with a vibrator-replacement inverter not starting was in a marine HF transceiver (conventional-valve TX, 12-volt-HT valves in the RX). The valve heaters were on all the time, TX/RX switching was done by switching +12V between the RX 'ht' line and the TX inverter-stage. The (initially unbiased - the entire design used grid-resistor bias) transmit-strip valves presented a very low-resistance load across the HT line which was enough to cause an occasional 'misfire' on the inverter when you pressed the transmit-switch and got no RF output.

I added an extra little 250V normally-open relay between the inverter and the transmit strip, powered from the inverter output which pulled-in once the HT line had got up to strength. A kludge, yes - but 100% reliable.

I nagree that Ge transistors are handy in these applications because of their low Vbe when saturated - particularly important in 6-volt applications!

[Bazz4CQJ]

Originally Posted by G6Tanuki

I nagree that Ge transistors are handy in these applications because of their low Vbe when saturated - particularly important in 6-volt applications!

So, keeping the inverters from my old Pye RT sets was a good move then ?
I am amazed just how efficient the original electromechanical vibrator was. I wonder how that performance may have changed as the vibrators aged?

B

[ex seismic]

And nowadays there are these:
http://www.royalsignals.org.uk/vibs/

Gordon
G7KNS

[G6Tanuki]

Originally Posted by Bazz4CQJ

So, keeping the inverters from my old Pye RT sets was a good move then?

Yep! I've got a few inverter- and modulator-NKT404s recovered from old Pye Vanguards and Cambridges; they may yet find their way into my R209s as vibrator-repalcements.

[pmmunro]

If anyone wants to keep their equipment original, I have an unused Plessey 12V non-synchronous vibrator on a B4 base looking for a home, and possibly a second one too.

PMM.

[rovernut]

Hi

Firstly thanks for sharing the information, I have only briefly looked at it but it will make very interesting reading.

Interesting about the comments about using back to back 2N3055, there is great variation of Betas on 2N3055’s (recent selection from reputable supplier ranged from 34 to 120). I have found that if the Betas of both are relatively close (say within about 5 of each other) they will oscillate, however if they are wildly different oscillation often doesn’t occur.

I have made several different types of inverter, including ones using a 555 timer to simulate the nominal switching frequency of the original vibrator, however I now tend to use the ones supplied by http://www.royalsignals.org.uk/vibs/ as noted above. The kits are of good quality and cheap, I have also used them in various Larkspur sets including the B47, B48, C13 also in R109 and R209’s I have attached some of the notes on the C13 (which uses a synchronous vibrator). I have also attached a few others on vibrator circuits that I have come across on the net. (I can't remember if I have built or tried the last two attachments)

Also have a look at these links
The Mallory Design Book

http://www.tubebooks.org/books/mallory_vibe.pdf

and a chapter by Rawlings on Vibrator Power Supplies

http://frank.yueksel.org/other/RCA/R...r-Supplies.pdf

73
Richard
M0YSR

[BottleMan]

When presented with an old car radio to repair that had a failed vibrator I read the Mallory book and also took in a fair bit of information from forum conversations etc.
I must say that, regardless of the "it seemed to work OK for me" observations of the performance of some of the basic designs, they do not generally observe good design practice and probably work in spite of themselves or because of some robustness in the rest of the design. Simply replacing the vibrator with a 50/50 power switching stage defeats the purpose of the resonant capacitors across the transformer. They are there to control the polarity reversal on the transformer during the period when the vibrator contacts are open circuit. As a conducting contact opens, the voltage rises in a sinusoidal fashion, reducing arcing. Depending on any damping that might be present, the other opposite phase of the winding progresses to a point where there is a low voltage across the next contact to make (the same as "zero voltage switching" in modern power supplies). To force the transformer/tuning capacitor into a condition of 50/50 switching creates unintended currents to flow in the windings. Therefore, might I suggest that those of you taking this approach might experiment with removing the capacitors. Alternatively, you could take the approach I did, leaving the existing circuitry in place and make an electronic vibrator using a bog standard push-pull PWM controller driving some MOSFETs with some dead-time programmed in. My circuit was not terribly optimised as I used what I had on the shelf, but to come up with something in the spirit of Mallory, you should only need the above and a sprinkling of passives.

Graham

[Argus25]

Replying to Bottleman's remark about 50/50 duty cycle causing unintended winding currents:

This is not correct. The transformer being inductive by nature opposes a change in current. So with a rapid step change in applied voltage the initial currents are very low. So when the transistor or fet switches, regardless of the voltage on the transformer terminals at that time, the instantaneous current, in the transformer windings at least, are very very low.

But the reverse is true for the tuning capacitors in parallel with the windings, these oppose a change in voltage, and the peak currents around those will be very high if they have not commutated to the same voltage as the new potential they are being switched to. But the peak current is limited by the impedance of the switching device.

So one could think as suggested that it would be a good idea when fitting an electronic switching method of a 50/50 duty cycle type to remove the tuning capacitors. However, this is not the case and likely if done the C-E voltage or D-S voltage of a mosfet would be exceeded. This is because the capacitance tunes the transformer's leakage inductance at switching (as noted in the article I posted) so reducing the cap values for most common vibrator transformers results in very large shorter duration (higher frequency) voltage spikes on the switching edges. (More modern bifilar wound transformers for DC/DC converters were much better in this respect)

In fact for any oscillator driven electronic version with a 50/50 duty its better to increase the the tuning cap value with additional capacitance to minimize the voltage overshoot as shown in the article.

I did use a signetics PWM IC in one version and created the timing of the electro-mechanical unit, nothing was gained here except a reduced output voltage. The 50/50 method is superior and much simpler, provided one pays attention to the tuning and is aware of the leakage reactance problem explained in the article.

Hugo.

[Argus25]

....I could also have mentioned on the post above, that there are three advantages of the self oscillating versions. 1) The rate of change of voltage on the transformer terminals is set by the existing tuning capacitors, so when the circuit changes state, the speed that one transistor comes out of conduction and the other goes into conduction is set by the transformer and its existing tuning caps. As noted in the article, this gives a much lower overshoot on the switching edges than when the transistors are driven by an external oscillator. 2)The self oscillating versions have fewer parts than a circuit with a separate oscillator and are therefore more elegant. 3) Due to the fact that the feedback is derived from the main transformer, the circuit is intrinsically short circuit/overload protected, if overloaded or shorted out the supply shuts down and draws no current. In the case of a separate oscillator driven version, the output devices simply get pulled out of saturation with overload and cook up, assuming a fuse doesn't save them.

[Oldcodger]

I had a play with a kit to replace the mech vibrator in the radio in my first car. From memory the kit came from one of the London stores and featured a self oscillating transformer arrangement with two 3055. I found it produced a large amount of noise into the HT line.
MSDS at one time produced a IC invertor regulated supply for the RT353 (Clansman),which produced a fair amount of power ( given that the TX stage at the low end could shove out 50W RF, WITH VOLTAGE ON ANODE AT 800V DC -manual states 600, but from measurement ,I can state it's nearer 800v), from a 24v SUPPLY.
Marconi also used the TL 494 IC in telecomms racks/shelves/cards to give a stabilised low noise supply to FDM kit.
Might give developers ideas for experimentation.

[Argus25]

Oldcodger,

Some of those kits could go unstable at high frequencies with 2N3055's unless snubber R-C networks were placed on the collector circuits to ground, or 0.1uF collector to base Miller capacitors added. Germanium power transistors seldom have this problem due to their lower transition frequencies.

There were a range of switching supplies to get high voltage in cars, some of which ran a higher frequency in the 400Hz to 3kHz region, these were notorious for generating interference. It is interesting that the higher the switching frequency, and to still maintain a square profile for the switching voltage, the rise and fall times have to be faster.
So the higher frequency Fourier components increase and radio interference is more likely. Folks who have tried to build switching supplies for getting the HT for battery valve radios have found this out.
One way to get around this problem is to go the other way, for the lowest possible operating frequency (40 to 60 Hz). The transformer is bigger, as are the smoothing caps, but if the transformer primary is tuned with a large value capacitor, the rise and fall times can be slowed to the extent that there is no component in the RF spectrum, at least above 100kHz. (Again this is another advantage of a self oscillating supply with adequate transformer tuning) Have a look at the end of this article at the switching waveform of the Omega Device, especially the sinusoidal like rise and fall:

http://www.worldphaco.com/uploads/WORLDFETRON.pdf

An LW or MW radio with a ferrite rod can be put directly on top of the Omega Device pcb, and it picks up nothing.