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 > Other Discussions > Homebrew Equipment

Notices

Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members.

Closed Thread
 
Thread Tools
Old 29th Jun 2009, 1:58 pm   #21
Station X
Moderator
 
Station X's Avatar
 
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,193
Default Re: Solid State Vibrator.

The CD4047 does appear to be a web only item from Maplin. I've ordered 5 from my RS account at the princely sum of 17P each post free. They'll be here tomorrow.
__________________
Graham. Forum Moderator

Reach for your meter before you reach for your soldering iron.
Station X is offline  
Old 29th Jun 2009, 2:54 pm   #22
mhennessy
Dekatron
 
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
Default Re: Solid State Vibrator.

Quote:
Originally Posted by jimmc101 View Post
There are a couple of points to be wary of using the transistor astable:
For a supply over about 5v base emitter reverse breakdown can occur when the bases are driven negative.
Yes, good point - solved with a pair of 1N4148's, but at the cost of component count...

Quote:
The rise time at the collectors is very slow since the cross coupling capacitors have to charge via the collector resistor.(The other end is clamped by the BE junction of the opposite transistor).
Less sure this is a real issue here, from memory it's more an issue of rounding the corners (?), which shouldn't cause significant switching loss in the MOSFET.

BTW, excellent link - wasted quite a bit of my lunch hour there!

But it's hard to argue at 17p for a CMOS IC! I do have a soft spot for transistor designs, partly nostalgic of course. Look forward to hearing how it goes tomorrow
mhennessy is offline  
Old 29th Jun 2009, 10:49 pm   #23
Station X
Moderator
 
Station X's Avatar
 
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,193
Default Re: Solid State Vibrator.

I think I'm getting closer to solving the overheating problem. First thing I did was replace the surface mount MOSFETs with TO220 versions. It's then dead easy to hold the heatsink tab between finger and thumb to detect overheating before smoke comes out. As the step up transformer was an unknown quantity I made up a dummy load consisting of two 12V 50W car headlight bulbs in parallel. With a 50% duty cycle and a 13.8V supply these would pull about 3.6 amps through a single MOSFET.

Powered up, connected dummy load to one MOSFET. Bulbs lit, MOSFET stayed cool and seemed willing to go on forever. I left it running for about 15 minutes after which time the MOSFET was still cool.

Connected dummy load to the other MOSFET. Bulbs lit, but within about 30 seconds the MOSFET's heatsink was too hot to touch. Switched off immediately.

With no load connected I used a scope to investigate the waveforms on the MOSFET gates. The cool MOSFET had a perfect square wave, but the waveform for the hot MOSFET rose sharply at first and then began to curve over to the right. The same thing happened on the falling edge. The waveforms on the outputs of the CD4011 were perfect square waves so it looked like the problem was in one of the 100K gate resistors. On measuring them I found that one was short circuit. It turned out that in soldering the gate lead of one of the new MOSFETs I'd managed to bridge the resistor's leads with solder. So I shorted the other one too. Both MOSFETs then ran cool when powering the dummy load for extended periods. I obviously need lower value gate resistors.

So at this stage both MOSFETs would drive dummy loads and there were perfect square waves on their gates and drains.

I then connected one MOSFET to one half of the step up transformer primary. The MOSFET stayed cool. Nice square wave on the gate, but much modified waveform at the drain due to the inductance of the transformer. Tried the same MOSFET it on the other half of the step up transformer. MOSFET was still cool, so it looked like the transformer was OK.

Tried to do the same thing with the other MOSFET, but it immediately overheated. It was however quite happy to power the dummy load while the other MOSFET powered the step up transformer. Obvously it didn't like an inductive load.

It then occurred to me that one MOSFET was on for 40% of the time and the other for 60% of the time. It was the one that was on for 60% of the time that didn't like the inductive load. At that point investigations had to cease. It will be interesting to see what happens when I have a proper 50/50 square wave to drive the MOSFETs.
__________________
Graham. Forum Moderator

Reach for your meter before you reach for your soldering iron.
Station X is offline  
Old 30th Jun 2009, 10:09 am   #24
mhennessy
Dekatron
 
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
Default Re: Solid State Vibrator.

You didn't mention 100K gate resistors before! 1K is better, although 0K is probably fine... When you said the gate waveforms were fine, presumably these were measured at the 4011 rather than directly on the MOSFET gates?

As this gave me cause to look again at the CMOS oscillator .DOC, I realised that you'd probably get closer to 50% if you used gates with Schmitt trigger inputs - in fact you should really be doing that anyway because of the "analogue" waveform being applied to one of the inputs. I know that your 4047 is due to arrive today, but if you have a 40106 (hex Schmitt inverter) handy, it's something to try. Certainly, last time I built one of these it gave 50%, although it was the simpler form of oscillator that only used a single gate. (Sorry for not thinking to suggest it before you ordered the 4047)

By the time I'm back in the office tonight, I expect to see a working design
mhennessy is offline  
Old 30th Jun 2009, 10:31 am   #25
Station X
Moderator
 
Station X's Avatar
 
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,193
Default Re: Solid State Vibrator.

Quote:
Originally Posted by mhennessy
When you said the gate waveforms were fine, presumably these were measured at the 4011 rather than directly on the MOSFET gates?
Quote:
Originally Posted by Station X
I am seeing clean square waves on both gate and drain. I haven't displayed them both at the same time though, but could do as the scope is dual beam.
That was with the 3055 series surface mount devices. It was only with the TO220 (HUF75337P3s) fitted that the waveforms became distorted. Different gate capacitance perhaps?

Quote:
Originally Posted by mhennessy
if you have a 40106 (hex Schmitt inverter) handy, it's something to try.
If I'm not careful this homebrew vibrator will start to suffer from what we called "creeping elegance" at work. In other words the law of diminishing returns comes into play. I know that my time costs nothing, but it might be better spent doing something more productive like restoring radios.
__________________
Graham. Forum Moderator

Reach for your meter before you reach for your soldering iron.
Station X is offline  
Old 30th Jun 2009, 11:22 pm   #26
Station X
Moderator
 
Station X's Avatar
 
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,193
Default Re: Solid State Vibrator.

Quote:
Originally Posted by mhennessy
By the time I'm back in the office tonight, I expect to see a working design
I'm afraid not. Partial success followed by total failure!

Wired up the CD4047. 1000pF capacitor and 2.2M resistor in the timng circuit. 220R gate resistors. MOSFETs not fitted at this stage. Checked it with the 'scope. Two beautiful square waves of opposite phases with a 50/50 mark space ratio.

Fitted MOSFETs and confirmed that there were nice square waves on the gates. Connected the dummy load to each MOSFET in turn. Bulbs lit, MOSFETs remained cool.

Removed (valve) rectifiers from the set. This meant that the solid state vibrator would only be connected to the step up transformer, the secondary of which would be isolated.. Tried each MOSFET in turn driving each half of the primary winding in turn. MOSFETs stayed cool. 100Hz hum from the transformer.

Connected both MOSFETs to the primary at the same time. No overheating. Let it run for about five minutes. MOSFETs still cool. Looked promising, this is further than I'd ever got before using MOSFETs. Obviously the 60/40 mark/space ratio of the earlier design had been causing problems. Measured about 1000V AC across the whole of the secondary winding.

Switched off. Isolated the smoothing and reservoir capacitors from the rest of the set by sleeving the contacts of the transmit/receive relay. Fitted the rectifier valves. Switched on again. Rectifiers warmed up. Measured 400V DC at the smoothing capacitor.

At this point I decided to load up the HT supply gradually using wire wound resistors. I left things running whilst I sorted out the resistors.

Suddenly the 100Hz hum ceased. Before I could switch off the 16A fuse (the correct specified value) in the set's 12V input blew. Investigation showed that the CD4047 was too hot to touch and had been destroyed along with the MOSFETs. The 13.8V 16A bench power supply had also taken a hit. The output voltage had dropped to 8V on no load.

First priority will be to fix the bench power supply, but I'll be using a car battery and a smaller fuse for any future testing. I don't know what went wrong in what order. I'm guessing that the oscillator stopped running leaving one or both MOSFETs switched hard on. The resistance of the step transformer's primary winding is only just over an ohm so heavy current would flow, but I wouldn't have expected it to reach the 32A or more needed to blow the fuse.

Four CD4047s and about 10 MOSFETs to go before I give up.
__________________
Graham. Forum Moderator

Reach for your meter before you reach for your soldering iron.
Station X is offline  
Old 30th Jun 2009, 11:31 pm   #27
AlanBeckett
Rest in Peace
 
Join Date: Dec 2008
Location: Burton upon Trent, East Staffordshire, UK.
Posts: 1,686
Default Re: Solid State Vibrator.

Graham,
You've got to remember that it's the Salesmen who bt about the products, the MD gets the credit but it's the designer who gets to turn it on!
Without wishing to encourage you to embelish the circuit, a free-running oscillator and a divide by 2 is much 'safer'. Also, if you capacitively couple the Gates, with a pull-down resistor, even if the oscillator stops it will be OK.
Alan
AlanBeckett is offline  
Old 30th Jun 2009, 11:53 pm   #28
Station X
Moderator
 
Station X's Avatar
 
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,193
Default Re: Solid State Vibrator.

The chip actually runs at 200Hz and has a built in divide by 2 down to 100Hz.

I take your point about AC coupling the chip to the gates, but I'm concerned that this may distort the square wave, which was the cause of previous problems.

I shalln't be giving up on this just yet, but I'll be taking a lot more precautions when undertaking future tests.
__________________
Graham. Forum Moderator

Reach for your meter before you reach for your soldering iron.
Station X is offline  
Old 1st Jul 2009, 5:20 pm   #29
mhennessy
Dekatron
 
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
Default Re: Solid State Vibrator.

Sounds like you had fun yesterday

As the IC got hot, it sounds like it went into SCR-latchup, probably caused by back-EMF in the primary finding its way through a MOSFET via some intrinsic diode or capacitive coupling. You mentioned that the drain waveforms were modified by the transformer (exactly what you'd expect, of course), and it's likely that as the load was increased, some critical voltage was reached...

Back in post #15 I mentioned my worries about using CMOS for this, although admittedly I didn't envisage this exact scenario. Assuming you want to persevere with the 4047, I'd suggest the following measures:
  1. Power the IC via a resistor - 1K to 10K (ish) should be fine - this will limit the current should the IC enter SCR-latchup again. A decoupling capacitor between the power pins is a good idea, but keep it small - perhaps only 100n or so... With this in place, the drive to the MOSFETs should disappear during a latchup, so hopefully they won't burn up. The circuit should start working again after a "reboot"...
  2. Increase the gate resistors as much as the switching characteristics allow - which will reduce the chance of a latchup...
  3. Connect 12V (ish) Zeners between gate and source to protect the gate channel.
  4. Include "catch" diodes between drains and +Vin and perhaps 0V too. These diodes might mean measures 1 and 2 aren't necessary.
  5. Perhaps not necessary, but the 2M2 sounds very high for such a circuit and the "non-driven" end might be picking up noise. Personally, for peace of mind, I'd be inclined to use 220K and 10n or perhaps even 22K and 100n (but I wouldn't want any more capacitance than that)

You might not need all of these, but I'd consider points 1&2 as essential. Unfortunately it means a few extra components, but that's essential real-world engineering - we're nowhere near your "creeping elegance" threshold yet

Once it's working again, put a scope on the gates to makes sure their capacitive load is being driven correctly, and check that the supply to the 4047 doesn't have significant switching ripple on it. Also carefully monitor the drains to see the voltage limits of the waveform under differing load conditions. Catch diodes (point 4) or maybe even a snubber network might be needed - bear in mind what happened to the contacts over time - that energy and dV/dT is still there!

Incidentally, the AC coupling could be made to work, and is used in commercial circuits, but it shouldn't be necessary once you've attended to the above...

Good luck,

Mark
mhennessy is offline  
Old 1st Jul 2009, 6:05 pm   #30
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
Default Re: Solid State Vibrator.

Just catching up on this post, I'd agree with everything Mark said except one! Don't have catch diodes from drains to V+. That is because, when one MOSFET turns on, by transformer action, the other's drain must rise to twice V+ and if you've got a catch diode it can't do that. Something will get hot, fast...

The gate resistors... yes 100k was far too high. It will slow down the switching (as you have already found out). 10 ohms is my own 'rule of thumb', but at the low frequencies you are using 220 ohms is fine. Solder the resistors as close as possible to the gate leads - they will help prevent parasitic oscillations, which can otherwise occur at many MHz.

The idea of AC coupling to the gates is inelegant. It is used 'out there', but far better to persevere with a reliable oscillator. If you do need to buy any components, an SG3524 (or equivalent) is the IC to go for - it's got oscillator, divide-by-two (for equal push-pull drive), variable duty cycle, MOSFET drive capability, all in a 16-pin package.
kalee20 is online now  
Old 1st Jul 2009, 6:15 pm   #31
AlanBeckett
Rest in Peace
 
Join Date: Dec 2008
Location: Burton upon Trent, East Staffordshire, UK.
Posts: 1,686
Default Re: Solid State Vibrator.

Re capacitive coupling, I would suggest that 'out there' has a lot more collective experience than 'in here'.
Alan
AlanBeckett is offline  
Old 1st Jul 2009, 6:25 pm   #32
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
Default Re: Solid State Vibrator.

No doubt Alan, if you can access it.

Luckily, I have 8 years experience with switch-mode power supplies, mainly MOSFETs but also bipolar, for a world-class power supply manufacturer.
kalee20 is online now  
Old 1st Jul 2009, 6:37 pm   #33
AlanBeckett
Rest in Peace
 
Join Date: Dec 2008
Location: Burton upon Trent, East Staffordshire, UK.
Posts: 1,686
Default Re: Solid State Vibrator.

No doubt, and I've had 40 years experience designing high reliability equipemnt, including PSUs, for several 'world-class' Companies.
It's axiomatic that oscillators are unreliable. It's their very instability that makes them work. I don't see any elegance in a design that destroys itself when something goes wrong.
Alan
AlanBeckett is offline  
Old 1st Jul 2009, 6:54 pm   #34
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
Default Re: Solid State Vibrator.

OK Alan, we're talking the same language (since the 8 years in purely power supplies, I've been 19 years in high-rel myself).

Relaxation oscillators such as the 3525 don't have a latch-up mode, and I'd consider them reliable. I can't speak for the 4047 as I've not used it. The classic two-transistor Eccles-Jordan multivibe does have a stable, non-oscillating state so I've avoided it always. And resonant oscillators can take ages to start so again I agree with you.

However, AC coupling also has the quirk that the first few cycles of gate drive will be different from the rest, until the coupling capacitor's charge stabilises. It also needs a DC-restoring diode gate-source, unless you are prepared to have less positive drive (and some drive going negative, which won't do any good).

I used the 3525 in a 450W, push-pull power supply, driving MOSFETs. Although only 45 were built, there has never been a FET failure in this unit, ever - and this includes both the bread-board prototype and the pre-production model! But I've blown many FETs getting the experience!
kalee20 is online now  
Old 1st Jul 2009, 7:24 pm   #35
Sean Williams
Dekatron
 
Sean Williams's Avatar
 
Join Date: Feb 2005
Location: St.Ippolyts, Hitchin, Hertfordshire QRA IO91UW
Posts: 3,517
Default Re: Solid State Vibrator.

Gentlemen, this isnt a forum for whos goes fastest! - lets keep this civil and on topic....
__________________
Engineers make things work and have spare bits when finished
Sean Williams is offline  
Old 1st Jul 2009, 7:58 pm   #36
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
Default Re: Solid State Vibrator.

Point taken, Sean.

Hopefully there's enough material above for Station X to see the next step forward, too!
kalee20 is online now  
Old 1st Jul 2009, 10:46 pm   #37
jimmc101
Heptode
 
jimmc101's Avatar
 
Join Date: Mar 2009
Location: Portsmouth, Hampshire, UK.
Posts: 674
Default Re: Solid State Vibrator.

Sorry to here of your problems.

As far as I am aware the only ways to cause a CMOS device to crowbar are:
Drive an input or output outside the supply rails so that more than 10mA flows.
Exceed the supply voltage rating.
Blast it with ionising radiation.

I would suggest feeding the 4047 via a resistor of a few hundred ohms and connecting a 12v zener across pins 7 & 14.
This will protect against supply over voltage.
Decouple with 100n + a good few uF to ensure Vdd remains clean.

Adding catch diodes from the Mosfet gates to ground and Vdd (pin14) should prevent more than 10mA flowing back into the 4047 outputs.
(via the 220 ohm gate resistors)

I don't think you need to bother with lead shielding.

Regarding Mosfet failure if inductive kick back was a problem then I would have expected the no (secondary) load to be the worst case.
All I can suggest is to monitor the drain voltage an add snubbers if spikes are too big.

One final thought with such a heavy pulsating load did your bench power supply loose regulation and over-volt?


Good luck

Jim
jimmc101 is offline  
Old 2nd Jul 2009, 9:40 pm   #38
Station X
Moderator
 
Station X's Avatar
 
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,193
Default Re: Solid State Vibrator.

I have repaired the bench power supply. Not having a circut for it, and not having the time to reverse engineer it either, I used the no brainer repair method. In other words I tested the transistors and replaced a faulty one, namely a twopenny BC337-25 on the control board. I photographed both sides of the PCB so I can reverse engineer it later. The PSU now delivers 13.8V at 16A and shuts down under overload or short circuit conditions. It comes back to life again if the overload/short is removed. I think the transistor has taken a high voltage hit, but the three 2N3055s in the output and the transistor driving their bases survived. This makes me think that using a pair of transistors as a multi-vibrator in the solid state vibrator might be no more resilient than a CMOS IC. I shall use a car battery for future experiments. It's bad enough having to rebuild the circuit under test without having to repair the test gear. I'm dreading blowing up the 'scope.

Turning my attention to the solid state vibrator I rebuilt it as before. With light bulbs as a dummy load it is happy to run all day without overheating and the gate waveforms are perfect. This makes me think I have a good circuit, but just need to make it more resilient by following the advice offered by contributors to this thread. I shall stick with the CD4047 until I get a resilient circuit or all the CD4047s have been destroyed. It seems to me that using light bulbs a current limiters for the drains might be a good idea when trying the circuit in anger. So long as they'll stand up to back EMF etc. that is.

It's interesting that all the circuits I've used so far are based on published designs, including one which is marketed as a kit, yet they don't work correctly.

I have thoroughly checked all the components in the power supply section of the set, including meggering the transformer to check for insulation breakdown at high voltage. Everything was in order.

I shall have to put this project to one side for a day or two, but will report any developments.
__________________
Graham. Forum Moderator

Reach for your meter before you reach for your soldering iron.
Station X is offline  
Old 3rd Jul 2009, 12:55 pm   #39
geofy
Retired Dormant Member
 
Join Date: Jun 2006
Posts: 1,798
Post Re: Solid State Vibrator.

Quote:
Originally Posted by kalee20 View Post
I used the 3525 in a 450W, push-pull power supply, driving MOSFETs.
This chip was used in the laser driver circuits I serviced, a surface mount version, though a standard version was also used in the power supply as a somewhat elaborate cooling fan driver, using the PWM as the speed control via a thermister on the laser block heatsink.

It has an op amp comparator internally linked to the PWM output MOSFET drivers, also a 5 Volt reference pin and enable and shutdown pins.

G
geofy is offline  
Old 3rd Jul 2009, 4:29 pm   #40
mhennessy
Dekatron
 
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
Default Re: Solid State Vibrator.

Quote:
Originally Posted by Station X View Post
This makes me think that using a pair of transistors as a multi-vibrator in the solid state vibrator might be no more resilient than a CMOS IC.
Glad you're back on track! Just wanted to comment on the above - I suspect that you've experienced a really common failure mode for regulated power supplies that have insufficient protection against load-dump, caused by the first transistor in the output stage (be it driver or pre-driver or whatever) suffering from excess reverse VBE - the emitter is taken positive by the external load, while the error amp is pulling it down because it's seen the output rise. We use lots of oldish Farnell PSUs (E30s), and these are easily damaged in experiments using motors or large inductors. I've fitted protection diodes to some...

Or in other words, it's not fair to draw any conclusions about the reliability of CMOS vs transistors based on this incident (and I promise that I dropped any pro-discrete "bias" while writing this)

BTW, if in the future you ever get around to reverse-engineering the circuit diagram of your PSU, please post it - it would be good to see if my supposition above is correct, and perhaps suggest how you might be able to add protection for the future)

Oh, and careful with that car battery
mhennessy is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 2:35 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.