UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > General Vintage Technology > Where To Get Sets and Parts

Notices

Where To Get Sets and Parts For discussions about swapmeets, rallies, NVCF and BVWS, car boot sales, antique and charity shops, dealers, newspaper adverts, the local tip and just about any other source of equipment (other than eBay).

Closed Thread
 
Thread Tools
Old 1st Aug 2018, 1:43 pm   #1
peter_scott
Dekatron
 
peter_scott's Avatar
 
Join Date: May 2003
Location: Edinburgh, UK.
Posts: 3,273
Default Vibrators Solid State Replacements

https://www.royalsignals.org.uk/vibs/
peter_scott is offline  
Old 1st Aug 2018, 10:13 pm   #2
HamishBoxer
Dekatron
 
HamishBoxer's Avatar
 
Join Date: Aug 2007
Location: W.Butterwick, near Doncaster UK.
Posts: 8,923
Default Re: Vibrators Solid State Replacements

And cheap enough to build your own if you wish or available ready built. Also a lot less "nosier" than the original I understand.
__________________
G8JET BVWS Archivist and Member V.M.A.R.S
HamishBoxer is offline  
Old 2nd Aug 2018, 12:16 am   #3
Argus25
No Longer a Member
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
Default Re: Vibrators Solid State Replacements

I think there is an issue that has not been addressed with the design (at least I cannot see it on the picture of the pcb). I'm surprised because the article I produced for making solid state vibrators for wireless sets, like the ZC1 (similar to the WS19) has been around for some years.

What is the issue ?

1) When a solid state vibrator circuit (which is force driven by an oscillator) switches the primary of a vibrator transformer, the existing value of the commutation capacitance (tuning capacitance value) on the transformer's primary is significantly inadequate and needs to be increased. Otherwise the voltage transients due to the transformer's leakage reactance threaten no only the switching devices, but also the transformer's insulation. The voltage transient magnitude will depend on the particular transformer and its existing tuning capacitances.

At least this is an easy fix, by placing a drain to drain capacitance of a suitable voltage rating and uF value, but I cannot see an added drain to drain capacitance in these units. This also reduces radio interference from the electronic unit, but it is fair to say any electronic version has much less of this than an electro-mechanical vibrator.

The relevant data on this issue is on pages 41 to 47 of this article, notice the text in RED on page 47. I made it like that so it would stand out and not get missed:

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

(Of note this issue does not apply to self oscillating circuit configurations)

Otherwise I think the physical design of the units look very good and nicely presented and probably better than any I have seen so far. The additional capacitance can always be added across the vibrator's socket pins.

(The voltage transients with a normally working electro-mechanical vibrator are actually lower than forced electronic switching because of the delay period where neither contact was closed, a dead band or timing gap. The resonant frequency of the transformer and tuning cap values chosen to allow good commutation with minimal overshoot. However, with mosfet and transistor switches, driven by an oscillator, with no significant dead band or timing gap used, the leakage reactance is a major issue, it is all in the article)
Argus25 is offline  
Old 2nd Aug 2018, 8:59 am   #4
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,060
Default Re: Vibrators Solid State Replacements

That doesn't seem right to me!

The 'dead time' with a mechanical vibrator, allows the transformer's primary inductance to gracefully resonate with a capacitor across the primary, so that by the time the vibrator's 'new' contacts have made, the voltage across them is already very low because a half-cycle of oscillation has taken place. So they join with minimal sparking. When the contacts separate, there is also minimal sparking because the primary capacitor prevents a sudden rise of voltage as they separate.

Replacing the vibrator with a solid-state version, the 'drop-in' replacement would ideally have a similar dead time. Switch-mode power supply IC's such as the SG3524 are easy to set up to provide suitable drive waveforms - the obvious thing is a preset pot, tweakable while 'scoping the transformer waveforms, so that the ideal condition is attained.

If the dead-time is nearly zero, then the time for half a cycle of free oscillation is reduced. So to get nearly lossless switching, the commutation capacitance has to be REDUCED, not increased.

Reducing the commutation capacitance has the unwanted side effect that spikes due to leakage inductance will be larger. (But this is a separate thing to consider).
kalee20 is online now  
Old 2nd Aug 2018, 11:40 am   #5
Argus25
No Longer a Member
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
Default Re: Vibrators Solid State Replacements

Quote:
Originally Posted by kalee20 View Post
That doesn't seem right to me!

The 'dead time' with a mechanical vibrator, allows the transformer's primary inductance to gracefully resonate with a capacitor across the primary, so that by the time the vibrator's 'new' contacts have made, the voltage across them is already very low because a half-cycle of oscillation has taken place. So they join with minimal sparking. When the contacts separate, there is also minimal sparking because the primary capacitor prevents a sudden rise of voltage as they separate.
I agree with these remarks and we are on the same page here. exactly.

However, there is more involved when there is no timing gap that allows this process.

When one of the mosfets goes into hard conduction, it switches one side of the primary winding across a fixed voltage (which is the power supply) at that moment, from the alternating current perspective, one half of the winding is effectively shorted out. This neutralizes all of the inductance in the other half of the primary, that is all of it that is magnetically linked.

The other half of the primary inductance proper electrically vanishes, leaving only the leakage inductance. This leakage inductance resonates with the winding capacitances, and the external tuning capacitances to create a very high Q circuit. As a result, shortly after one mosfet goes into conduction, a large voltage transient appears on the drain of the contralateral mosfet. The negative going component of it gets clamped off by the D-S parasitic diode in the Moffett.

Vibrator transformers were also not specifically designed for low leakage inductance between the two halves of the primary either, making it worse than usual.

Have a look at the waveforms it the article on the pages I cited.

Here is the real catch 22:

Most designers of vibrator substitutes think they have done a good job, because when they are finished they put the scope on the drain or collector and look at the waveform and it looks like a respectable square wave.

However, on a usual scope display, set to view about 1/2 to a few cycles of oscillation, it is very difficult to see these brief transients. They need to be expanded out and viewed on a good scope with a delay timebase. Have a look at the oscillograms in the article and you will see what I mean.

Hugo.
Argus25 is offline  
Old 2nd Aug 2018, 12:04 pm   #6
David G4EBT
Dekatron
 
David G4EBT's Avatar
 
Join Date: Feb 2003
Location: Cottingham, East Yorkshire, UK.
Posts: 5,737
Default Re: Vibrators Solid State Replacements

There have been a couple of threads on the forum in recent years on this topic, which might be of interest:

https://www.vintage-radio.net/forum/...ight=Vibrators

https://www.vintage-radio.net/forum/...t=41944&page=3
__________________
David.
BVWS Member.
G-QRP Club member 1339.
David G4EBT is online now  
Old 2nd Aug 2018, 4:04 pm   #7
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,060
Default Re: Vibrators Solid State Replacements

Quote:
Originally Posted by Argus25 View Post
Have a look at the waveforms it the article on the pages I cited.
I had done. Hugo thinking is as meticulous as Hugo handiwork. Rarely equalled, even more seldom bettered. Homework was therefore done before querying!

Quote:
Originally Posted by Argus25 View Post
...The other half of the primary inductance proper electrically vanishes, leaving only the leakage inductance. This leakage inductance resonates with the winding capacitances, and the external tuning capacitances to create a very high Q circuit. As a result, shortly after one mosfet goes into conduction, a large voltage transient appears on the drain of the contralateral mosfet. The negative going component of it gets clamped off by the D-S parasitic diode in the Moffett.
Yes, the leakage inductance is all that is left. The nice elegant resonance, with the resonating capacitor, is due to the magnetising inductance of the transformer and I'm exactly with you there.

The leakage inductance is all that is left when one of the vibrator contacts is 'made' or the equivalent mosfet is 'on'. The contralateral MOSFET, however, just before being it was turned 'off' is carrying load current as well as magnetising current, and this is flowing in the leakage inductance. When the device turns 'off' (or contacts separate), you get a big voltage overshoot which peaks when all the energy stored as current in leakage inductance is transferred to voltage across circuit capacitances. The ringing frequency is fairly high, because leakage inductance is much less than magnetising inductance, so things are easily missed, till the MOSFET dies.

As you say, you also get a similar ringing when the other transistor turns 'on' because the coupled turns of the half-primary are whacked up to double the supply voltage, leaving the uncoupled bit of leakage inductance with a step voltage change imposed across it - so it oscillates again.

Quote:
Originally Posted by Argus25 View Post
Most designers of vibrator substitutes think they have done a good job, because when they are finished they put the scope on the drain or collector and look at the waveform and it looks like a respectable square wave.

However, on a usual scope display, set to view about 1/2 to a few cycles of oscillation, it is very difficult to see these brief transients. They need to be expanded out and viewed on a good scope with a delay timebase.
A 'scope with delayed timebase is one of my 'must-have' instruments. Examining switching edges has formed a big part of my life for many years!

Your article deals with fettling an existing vibrator for symmetric operation; separating of opening and closing times for the twin pairs of contacts in a synchronous vibrator, etc. So having emphasized the importance of this, I'd certainly say that a vibrator replacement should guarantee equivalent precision - hence recommending the use of an SMPS control chip (running at an unusually low frequency!) for stability, equality of drive to each side, and easy adjustment of dead time.
kalee20 is online now  
Old 2nd Aug 2018, 9:39 pm   #8
Argus25
No Longer a Member
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
Default Re: Vibrators Solid State Replacements

Kalee20,

Yes in essence the high voltage transient is caused by a rapid rising edge 24V potential (the 12V supply + 12V stepped up by the other half of the primary) shock exciting a resonant circuit composed of the primary leakage inductance and the associated capacitances. The Q is high as the resistances are relatively low. As noted this happens to a large extent only in a force or osc driven circuit, not self oscillating(see below). I do agree that the reduction in load current in the leakage inductance plays a part, however on testing I found that the voltage transient magnitude was relatively independent of that, and a nearly identical transient is generated with negligible load. It is primarily because a new large step voltage is applied to the capacitances, rather than the energy stored in the leakage inductance's field generating the spike. As noted from the recordings the transient is only about 10uS wide at its base or half cycle time.

In a modern SMPS, normally, the primary winding is bifilar wound, minimizing the leakage inductance which helps. But these old vibrator transformers were not wound like that, its easy to tell, just by measuring the DC resistances, larger on one half of the primary winding than the other.

In the case where the circuit is configured to self oscillate, switching timing of the two devices is such that it takes longer for the device coming out of conduction to do that.This controls the transient and the changeover between one mosfet (or transistor) going into conduction and one coming out of conduction is more gradual, but probably less efficient. But it doesn't matter because the switching events per unit time are only in the 100 to 200Hz vicinity so the energy dissipation on account of this is low. So I preferred this method to others and the circuitry more robust than an IC driver. But I do agree an SMPS IC , like an SG3524, can be used to replicate the vibrator contact timing. Self oscillating types are intrinsically short circuit protected too.

Last edited by Argus25; 2nd Aug 2018 at 10:06 pm. Reason: typo
Argus25 is offline  
Old 2nd Aug 2018, 11:30 pm   #9
Argus25
No Longer a Member
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
Default Re: Vibrators Solid State Replacements

Kalee20,

I was thinking it would be good to design an experiment to assess what proportion of the voltage transient is created by the load current storing energy in the leakage inductance , vs the transformed voltage shock exciting the leakage inductance's resonant circuit.

See attached circuit (leakage inductances and tuning capacitances not shown). Simply by turning it into a single ended circuit and having no initial current in the leakage inductance and therefore no stored energy in it, before the mosfet turns on. It could be interesting.

What do you think will happen ? I'm going to try it in a while, but I'm away for a bit. It could be setup in spice for a transient analysis by inserting some values for the leakage inductance, resistance and capacitances.
Attached Thumbnails
Click image for larger version

Name:	leak1.jpg
Views:	160
Size:	27.4 KB
ID:	166967  
Argus25 is offline  
Old 3rd Aug 2018, 10:08 am   #10
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,060
Default Re: Vibrators Solid State Replacements

As drawn, the waveform at the RHS will be exactly a square wave, no ringing (assuming perfect coupling between the two halves of the winding, and a 1:1 duty cycle drive.

At the moment of the MOSFET turning 'on' the waveform will be just about to collapse anyway. 12V across the winding for a defined period will charge the core, followed by 12V in the opposite direction (assuming the Zener diode us exactly twice the supply voltage) will exactly discharge it again.

If however on the RHS you add a bit of downstream, uncoupled inductance, and stray capacitance, you'll get approximately the waveform you have drawn. In this case, when the MOSFET is turned 'on' for the first time, the extra LC circuit will be shock excited. One side will be jerked up from 12V to 24V; the other side will ring with 12V peak amplitude, which is likely with 'realistic' losses to die out by the time the MOSFET is asked to change state.

When the MOSFET is turned 'off' the whole thing will ring. If the coupled winding has very small inductance compared with the leakage inductance (an unlikely condition in practice), it will largely drive the situation and the leakage will be jerked down from 24V to zero. So you'll get 24V peak amplitude of ringing. And then 24V in the first direction... ad infinitum.

If however, the coupled winding has inductance which is not small - and in practice it is likely to be larger than the added leakage - then the whole thing will ring at a considerably lower frequency on MOSFET turn-off. I'm on a local train right now so can't sketch anything out, worse luck (picture being worth 1,000 words and all that) but hopefully you can follow the above?
kalee20 is online now  
Old 3rd Aug 2018, 10:22 am   #11
Argus25
No Longer a Member
 
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
Default Re: Vibrators Solid State Replacements

Kalee20, you are bang on with your analysis. In fact it is easy to see from the recordings I made that the energy source for the high voltage spike is in fact stored magnetic field energy in the leakage inductance, not transformed energy from the other half of the primary.

Two things give the game away. Firstly the DC axis of the transient (even though the negative half is clamped out) is close to zero volts. If it was transformed from a voltage step it should be around 24V.

As can be seen also, the voltage on the drain of the recently non conducting fet rises slowly after that of course, because of the limited frequency response and hysteresis of the iron cored transformer ! (It might be a different case if it was ferrite).
So clearly, the transient is powered by stored magnetic energy in the leakage inductance, exactly as you said
Argus25 is offline  
Old 7th Aug 2018, 6:39 pm   #12
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,060
Default Re: Vibrators Solid State Replacements

Thanks Hugo!

Reading through with a bit more leisure, your vibrator-replacement using the two AZX17 transistors... I'm wondering why you didn't want the drive transformer to saturate?

If you let the main transformer saturate, the collector current shoots up at onset of saturation, and the fixed base current is no longer enough to keep the relevant ASZ17 in full conduction. So the collector voltage rises, the voltage across the main and feedback transformer falls, the base current falls, and the whole thing changes state. But, it is preceded by a rather stressful collector current spike.

If you design the feedback transformer so that this saturates FIRST, then at the onset of saturation the primary current in the feedback transformer starts to rise. But, it is limited by the primary resistance of 144 ohms, so you don't get a massive collector current spike. With more voltage dropped across the winding's resistance there is less available magnetically - the voltage across the feedback coil drops, the ASZ17 starts to turn off, and the circuit changes state as before, but much more kindly to the ASZ17's.

It's good to see, however, that the original power transformer gave operation at 60Hz-ish. Normally, operating frequency (around 100Hz) would be well-defined by the natural frequency of vibrating reed with a little mass on the unsupported end. So at 100Hz, the original transformer would be comfortably away from saturation.

Last edited by kalee20; 7th Aug 2018 at 6:42 pm. Reason: Corrected type number of transistor!
kalee20 is online now  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 7:09 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.