4th Jul 2009, 11:14 pm | #41 |
Moderator
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,289
|
Re: Solid State Vibrator.
I rebuilt the circuit completely as the Veroboard was getting a bit tatty. Used a holder for the IC so that it could easily be changed in case of accidents.
2200uF Capacitor across the 13.8V supply. 1K Resistor to CD4047 pin 14. 12V Zener Diode and 100nF Capacitor across CD4047 pins 7 and 14. 22K Resistor and 100nF Capacitor to set frequency at 100Hz. 220R Resistors to MOSFET gates. 12V Zener Diodes between gates and sources of MOSFETs. I didn't fit catch diodes, but could do later. Connected to bench power supply. Pefect waveforms on gates. Connected headlamp bulbs as dummy loads. Perfect waveforms on drains. Bulbs lit. MOSFETs cool to touch. Connected car battery to rig. Powered unit from vibrator socket on rig, but did not make connection to step up transformer. Perfect waveforms on gates. The next few steps took a lot of courage!! Rectifiers had already been removed from rig. Made connections to step up transformer. Switched on. CD4047 and MOSFETs survived. 100 Hz hum from transformer. Perfect waveforms on gates. Waveforms on drains were much improved compared to previous attempts. About 1000V AC across the transformer secondary. Left things running for 15 minutes monitoring waveforms and current drawn. Switched rig on and off several times, but could not kill unit. Oscillator always started reliably. Isolated rig PSU by sleeving send/receive relay contacts. Inserted rectifiers. Everything still OK. About 600V DC at smoothing capacitor. Switched on and off several times. Everything survived and kept working. Connected 6.66K (3 X 20K WW Resistors) across smoothing capacitor with milliammeter in series. 257V, 39mA on LOW. 412V 62mA on HIGH. 1.2A drawn from battery on LOW and 2.7A on HIGH. Efficency 60% on LOW and 68% on HIGH. Actually it isn't that good, as the rectifier heaters draw 0.6A between them. I don't want to fit semiconductor diodes in their place. Switched on and off many times over a half hour period. All OK. So at last I have a working circuit. Thanks to everyone who contributed to this thread for their advice. I couldn't have done it without you. I still have a bit of work to do like rapidly loading and unloading the output to simulate morse keying. Also increasing the gate resistor values as much as I can without distorting the waveforms on the gates and determing the optimum frequency for the oscillator to suit the transformer and the capacitors across its secondary. I'm not sure how to tackle the latter. The next stage will be to document everything so that forum member Colin can produce a PCB for the production version.
__________________
Graham. Forum Moderator Reach for your meter before you reach for your soldering iron. |
4th Jul 2009, 11:41 pm | #42 |
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,088
|
Re: Solid State Vibrator.
Hi Graham,
Congratulations! Answering your questions: Increasing the gate resistors will slow down the switching, which will increase losses but reduce RFI (although in practice it doesn't reduce much). Frequency: I'd aim for a frequency close to the original vibrator's frequency. The transformer will be designed around this. The lower the frequency, the lower the switching losses (because you've got fewer switchings per second), but at some point you'll hit the saturation limit of the transformer, the current will get peaky, and it's bad news all round. Finally, the secondary capacitor! Hopefully you've got about 10% 'dead time' between one MOSFET turning off and the other turning on. You need to 'scope one of the FET drain waveforms. The idea of the capacitor is to resonate with the transformer's inductance during this dead time, so that when the next FET switches on, the drain voltage has already swung pretty low. This makes the losses as small as possible (with the original vibrator, it would minimise sparking at the contacts). If the FET didn't switch on, the drain voltage would continue to ring up and down as a train of damped oscillations. If you 'scope the darin, you can see this happening and you need to play with the capacitor value until it's right. It is best done empirically, as the transformer inductance tends to be level-dependent anyway. Too small and the drain voltage will have reached its trough and will be on its way back up again when the FET switches on; too big and the voltage will not have had chance to reach its minimum when the FET turns on. And the capacitor needs to be a good-quality high-voltage type, polypropylene probably, as it works fairly hard. Have fun - but be prepared to blow a few FET's. If you don't, give yourself a good pat on the back. If you have 2 power supplies, use one for the gate drive oscillator (at normal voltage) and the other for the power circuitry, so that you can wind it up from zero with full gate drive applied while you watch the 'scope. This minimises the chance of fireworks! |
5th Jul 2009, 11:46 pm | #43 |
Retired Dormant Member
Join Date: Mar 2006
Location: Sheffield, South Yorkshire, UK.
Posts: 1,587
|
Re: Solid State Vibrator.
What a great thread....
Mike |
6th Jul 2009, 11:55 am | #44 |
Retired Dormant Member
Join Date: Mar 2007
Location: Skelleftea, Sweden
Posts: 250
|
Re: Solid State Vibrator.
indeed, many good tips here.
|
7th Jul 2009, 7:07 pm | #45 |
Dekatron
Join Date: Nov 2006
Location: London, UK.
Posts: 3,496
|
Re: Solid State Vibrator.
Hi Graham,
I've been watching this thread with interest. I see that you need quite a modest output current and wondered if you had considered a simpler solution - buying one from Kosbo? A DC-DC design based on a single MAX1771 with one MOSFET and a linear voltage regulator can output 250mA from a 9V supply. These are marketed as NIXIE tube power supplies. The component count is tiny and the footprint is minuscule (3 by 4 cm, from memory) if you buy a prefabricated unit. You simply need to change one resistor to adjust your output voltage. I only mention it because they have an efficiency of 86%, which is hard to beat, and probably a factor for most of us.
__________________
Al |
7th Jul 2009, 7:35 pm | #46 |
Moderator
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,289
|
Re: Solid State Vibrator.
I think you're missing the point. I want a plug in replacement for a mechanical vibrator, which externally is indistinguisable from the original. These can be bought off the shelf, although they're quite expensive. That's why I decided to make my own.
I'm well satisfied with the design I now have and just need to build a production version. If I went for your solution I'd have problems like having to operate from a 13.8V supply and the need to switch between low and high voltage output.
__________________
Graham. Forum Moderator Reach for your meter before you reach for your soldering iron. |
7th Jul 2009, 8:02 pm | #47 |
Dekatron
Join Date: Nov 2006
Location: London, UK.
Posts: 3,496
|
Re: Solid State Vibrator.
Excellent.
__________________
Al |
7th Jul 2009, 10:08 pm | #48 |
Moderator
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,289
|
Re: Solid State Vibrator.
Astral Highway. Sorry if I was a bit dismissive of your suggestion to use a Kosbo. It came rather late in the day for me though as I've "frozen" the design I intend to use. I also have enough components available to build several examples, although I only need one.
However if anyone else is contemplating building an SSV (Solid State Vibrator) the Kosbo might be a good starting point as it provides a convenient way of getting hold of most of the components and a circuit board. The tiny transformer indicates that the unit operates at a frequency of several KHz. The transformer would need to be removed and components changed to drop the frequency to the 80 to 120 Hz range used by mechanical vibrators. The MOSFETs could then be connected to the set's step up transformer primary. Additional protection would probably be required to stop the unit destroying itself as did my early attempts. Now I must get down to documenting and building my design or forum members will think I'm just an armchair restorer.
__________________
Graham. Forum Moderator Reach for your meter before you reach for your soldering iron. |
7th Jul 2009, 10:20 pm | #49 |
Retired Dormant Member
Join Date: Mar 2006
Location: Sheffield, South Yorkshire, UK.
Posts: 1,587
|
Re: Solid State Vibrator.
Graham,
You are definitely not an armchair restorer... Mike |
7th Jul 2009, 10:38 pm | #50 |
Heptode
Join Date: Jun 2004
Location: Ilkley, West Yorkshire, UK.
Posts: 656
|
Re: Solid State Vibrator.
Nah! ... I think he does radios
Excellent thread by the way, I've been following it with great interest, and learned a lot! Waiting for the next installment is a bit like looking forward to 'Corrie' ... only much more entertaining!
__________________
Colin Armstrong |
7th Jul 2009, 10:58 pm | #51 | |||
Dekatron
Join Date: Nov 2006
Location: London, UK.
Posts: 3,496
|
Re: Solid State Vibrator.
Quote:
Quote:
Quote:
Best of luck with the rest of your project, Graham and I look forward to reading the sequel.
__________________
Al |
|||
10th Jul 2009, 8:57 pm | #52 |
Triode
Join Date: Mar 2007
Location: Round Lake Park, IL. USA
Posts: 30
|
Re: Solid State Vibrator.
Some time ago I designed a solid state power supply for my Paraset transceiver.
It runs on 6 Volts. I ran a test against a conventional vibrator type power supplywith over the counter parts. I could post the findings if any one is interested. Here' the circuit.
__________________
Valves....The other Integrated Circuit |
11th Jul 2009, 9:05 pm | #53 |
Dekatron
Join Date: Jan 2004
Location: Newcastle upon Tyne, Tyne & Wear, UK.
Posts: 8,195
|
Re: Solid State Vibrator.
Hi Graham, another tip:
The gate resistors will slow up turn ON and OFF. Add a small doide across them (or a diode and a lower value resistor). This will allow a much faster turn off and reduce stresses and dissapation in the MOSFETS, it will also improve efficiency a little. Ed |
12th Jul 2009, 3:16 pm | #54 |
Moderator
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,289
|
Re: Solid State Vibrator.
I've done a bit more on this. I'm 90% certain that the original mechanical vibrator used in the set was an "Oak" made by AWA in Australia. It lacks a paper label, but the internal construction is exactly the same and the metal can very similar. The designation V6273 is of the same format as other Oak models, but I've not been able to find any data for it. All Oak vibrators seem to have operated at 100Hz and this is the frequency used by the solid state version. However back in post 9 I measured the frequency of the mechanical vibrator as 83 Hz. Possibly it was maladjusted. Only one resistor would need changing to change the SS version's frequency to 83 Hz.
http://cool386.tripod.com/msp/msp.html As an aside vibrators such as the "Mallory" made in the USA seem to have operated at 115 Hz as do off the shelf SS replacements. This frequency + or - 8 Hz was recommended in a standard published in 1948. My set was made some time before this though. I have found that increasing the value of the gate resistors from their present value of 220R causes the MOSFETs to heat up.If 100K resistors are used as specified for the Tony Maher battery eliminator the MOSFETs get too hot to touch and would probably destroy themselves if the unit was left switched on for any length of time. It's said that 100K resitors were specified so as to give some "Off Time" between one MOSFET switching on and the other switching off. They do this in conjunction with the gate to source capacitance of each Mosfet by slowing the rise and fall times of the square wave which is fed to each MOSFET. This doesn't seem to work though. With a larger resistor the drain waveform slopes to the right on the rising edge and then curves over at the top, the reverse being true on the falling edge. I assume that this results in the MOSFET being only partially switched on for part of the cycle and acting like a resistor which turns current into heat. I'll be keeping the 220R gate resistors. This brings us on to the vexed subject of Off Time. Mechanical Vibrators had Off Time between one set of contacts making and the other breaking. It's inherent in their design. Current cannot be instantly reversed. Off Time allows the current to drop as the contacts change over and reduces sparking at the contacts. Off time also reduces spikes on the supply, the transformer and buffer capacitor forming a tuned circuit at the vibrator's frequency. This increases efficency and reduces interference. SS vibrators don't have contacts, so Off Time isn't needed to protect them. Scoping the waveform at the drains doesn't show any spikes as such. There is some overshoot as the voltage rises and then a decaying oscillation of short duration. Varying the vibrator's frequency from about 70 to 130 Hz made no difference whatever to this. There is some interference generated on MW and LW, but I would expect this, as the device is connected to the vibrator socket by long leads and is unscreened. The set isn't in its steel cabinet either. At this time I don't intend modifying the design to incorporate Off Time, but may have to do so if interference is a problem. The design described in post 41 is working well and I haven't managed to destroy it yet. A kind forum member has designed a PCB for it. When I have a PCB I'll build a unit and fit it into the can. Then once I've restored the set I'll be able to see how well it performs and whether it generates interference. Plenty there for you to discuss I think.
__________________
Graham. Forum Moderator Reach for your meter before you reach for your soldering iron. |
12th Jul 2009, 9:31 pm | #55 |
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,088
|
Re: Solid State Vibrator.
Increasing the gate resistors will affect the switching times as you say, as the MOSFET gate has significant capacitance. The trouble is, it's not particularly well controlled. If there's a slight misalignment of the masks used to make the device, the gate can overlap the source or the drain, increasing the capacitance.
Designing a circuit which relies on a specific value of capacitance for correct operation is a bit like designing a bipolar transistor circuit around a particular value of hfe - it's bad design. So, definitely keep your small value of gate resistors. Then, whatever the value of input capacitance, the switching times will be small in comparison with the 'on' and 'off' times. |
13th Jul 2009, 8:41 pm | #56 |
Moderator
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,289
|
Re: Solid State Vibrator.
Here's the circuit and artwork for the PCB courtesy of forum member Colin. The MOSFETs used were HUF75337P3s.
Board size is 1.2" by 2.6"
__________________
Graham. Forum Moderator Reach for your meter before you reach for your soldering iron. |
13th Jul 2009, 9:11 pm | #57 |
Retired Dormant Member
Join Date: Mar 2007
Location: Skelleftea, Sweden
Posts: 250
|
Re: Solid State Vibrator.
wow, quite some work put into that.
|
30th Jul 2009, 9:44 pm | #58 |
Moderator
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,289
|
Re: Solid State Vibrator.
I feel a bit guilty about the fact that Colin pulled out all the stops and made a couple of PCBs for me but I didn't get round to assembling a complete unit until last weekend. Here it is.
There were no real difficulties in assembly. The 2200uF capacitor had to be mounted on the other side of the "chassis" to enable the complete unit to be slid into the existing can. This meant soldering stiff wires to the copper side of the PCB, passing them through holes in the chassis and soldering the capacitor to them. Anyone building a similar unit could mount this cap on the correct side of the PCB if space wasn't at a premium. I decided not to bolt the MOSFETs to the PCB. I figured that if a fault caused them to overheat they'd be destroyed and I didn't want them to destroy the PCB as well. There's a slight gap between them and the PCB. I should have mounted the chassis to the base the other way round to avoid having so many of the wires to the pins having to pass through holes in the chassis. I'll know better next time. Unfortunately the holes in the base and chassis are not symmetrical, as they have to avoid the pins, so I can't just turn it round. Soldering the wires to the pins wasn't easy. I ended up packing the holes through the pins with extra pieces of wire to give a capillary gap for the solder to run into. Next time I'll sleeve down the pins. It all works fine. Full report with waveform pictures to follow shortly.
__________________
Graham. Forum Moderator Reach for your meter before you reach for your soldering iron. |
30th Jul 2009, 10:21 pm | #59 |
Moderator
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,289
|
Re: Solid State Vibrator.
When loaded to give a current of 60mA the HT voltage is 257V on low power and 414V on high power. Power switching is accomplished by adding extra turns to the secondary winding of the step up transformer. As the set uses valve rectifiers it's nice to watch the HT slowly rise as they warm up.
The gate waveforms look excellent with a 50% duty cycle and a frequency of 100Hz. There is some overshoot of the gate voltage, but this is quickly damped out. I haven't 'scoped the transformer secondary because of the high voltages involved. Note that there is no "OFF" time between one MOSFET switching off and the other switching on. OFF time is inevitable where a mechanical vibrator is concerned and gives the primary voltage time to decay before the vibrator contacts make again. This reduces sparking at the contacts. Solid state vibrators do not have contacts, so this does not apply. However it is also said that OFF time also reduces spikes and radiation of RF interference. The secondary of the step up transformer should be tuned to the vibrator's frequency by a buffer capacitor to reduce spikes. I have not checked this, but the units frequency of 100Hz is the same as the specified frequency of the "Oak" unit it replaces. I am planning to build a solid state unit with OFF time using a UC3525 Pulse Width Modulator IC so that I can compare the operation of the two units.
__________________
Graham. Forum Moderator Reach for your meter before you reach for your soldering iron. |
31st Jul 2009, 12:30 am | #60 |
Rest in Peace
Join Date: Dec 2008
Location: Burton upon Trent, East Staffordshire, UK.
Posts: 1,686
|
Re: Solid State Vibrator.
Graham,
As we used to say, that's extremely swish Alan |