Programmable push pull FET driver.
Used to replace vintage radio mechanical vibrators.
Embedded Files
22 April 2017 a new higher powered unit, up to 200VA at 24VDC input.
The above image shows the new PCB operating with minimum heat-sink, the output voltage as read on the scope is 194 VDC across a 577 ohm load ( P=V^2/R so 194^2/577= 65 watts). The 50 Hz square wave is a bit brutal on the transformer as the flux density is higher than normal, the input voltage is slightly under 12 VDC, running at 60Hz is a much better option. The dip switches seem to be set for 5% off time. This could be a reasonable B+ supply for a 12 volt tube amplifier, I would not run the MM2015 transformer over about 75 watts and would heat-sink the FET's. The MM2015 has quite high leakage inductance.
The microprocessor controlled mechanical vibrator replacement circuit board I designed in 2016 has been modified; the main changes are listed below.
1 The PCB size has changed from 2.5 cm x 5 cm to 2.94 cm x 6 cm.
2 TO-220 FET’s are now used enabling higher power and a larger choice of output devices.
3 Two new programs are available.
4 Three different FET mounting options are available.
5 The brown-out detection is much improved.
6 The voltage select DIP switch was removed allowing more DIP switch select-able routines to run. 50 Hz or 100 Hz with off times of 5%, 10%, 20% and 30%. Mechanical vibrators normally had an off time between 10 and 30%.
7 Pads for snubber networks have been added, this plus avalanche operation of the FET’s controls the back EMF.
8 Can be configured for up to 10 amps at 24 volts input.
9 The DIP switch settings can be changed on the fly.
Some Specifications:
The micro has no crystal fitted and relies on its internal oscillator, the 12F1822 data sheet (DS41413C) page 3 states the internal oscillator is factory calibrated to +-1%
Some graphs of oscillator characteristics can be found around page 385.
I will quote +-5% for this device, none tested have exceeded +-1.5%.
The chips programmable brownout is set at ~2.7 volts.
The 12F1822 Micro data sheet can be found here: http://ww1.microchip.com/downloads/en/DeviceDoc/40001413E.pdf 433 pages on a device 4 mm x 5 mm !
The FETs used are:
FET used for the 12 volt version:
NXP PSMN3R4-30PL N-channel MOSFET, 100 A, 30 V, 114W, TO-220
RDSon @ 4.5 Vgs 4 milliohm; maximum gate threshold 2.15V
FET used for the 24 volt version:
Infineon IPP037N06L3 G N-channel MOSFET, 90 A, 60 V, 167W, TO-220. Note: This FET is 3 times the price of the 30V device, more expensive devices can handle much more current.
RDSon @ <4 milliohm; maximum gate threshold 2.2 V
Input Voltage:
For the 12 volt version 6 to 14.5 V, if the blank pads for R4 (0805) are shorted together an input voltage of <5 to 14.5 with an increased quiescent current of the micro/drive circuit to 43 mA at 14.5 V in.
For the 24 volt version 10.5 to 28V, if the blank pads for R4 (0805) are shorted together an input voltage of <9 to 28V with an increased quiescent current of the micro/drive circuit to 11 mA at 28 V in.
The main dropping resistor is a 2512 type in series with a 1206 type that can be shorted out.
Maximum current : Any configuration 10 amps, normally a heat-sink is used at 10 amps. At < 5A heat-sinking has never been required for any tests so far.
Maximum voltage: Depending on version normally 14 or 28 volts. If the PCB is loaded with suitable components a 48 volt version or higher is quite possible.
Important Information.
The device has no built in current limit so a fuse or self resetting polyswitch should be fitted to the center tap in the transformer primary, do not try to discharge a large low ESR capacitor with the PCB!
Looking at the test with the Jaycar MM2015 100VA transformer the primary DC resistance is 160 milliohms add to this the FET RDSon and a fuse; the primary resistance is >200 milliohms, ohms law I=V/R gives 12/.2=60 amps is the maximum possible current, the FET will handle this easily as its rated current is 100A a single 8 amp fast fuse should suffice for protection. At 60 amps the fuse should operate in 100 milliseconds. A fuse or poly switch will add about 10 milliohms to the circuit resistance.
The 12F1822 micro will drive the FET's into hard conduction even at the point of brown-out shutdown, the FET's will not cook even if the battery slowly goes flat.
If different FET's are used they must have digital level gate drive capabilities and be on at the brown-out voltage.
Due to the current firmware a 1 second delay occurs at startup or after a brown-out reset, this prevents hi speed hiccups; recovery from brown-out is smooth when the supply voltage rises. I may if the need arises edit the firmware for a PWM soft start.
Possible uses:
1 New 12 or 24 volt tube equipment.
2 12 or 24 volt power supply for 240V LED lamps.
4 Restoring old valve radios and vintage military tube equipment.
5 Xenon flash lamp power supply.
6 Amateur radio projects.
Above is an image of the dual clip mounting option. The spring clips are RS part number 504-0592. Using spring clips is the preferred mounting method for low thermal resistance, some silicon thermal pad or a mica is used between the FET’s and heatsink, a small piece of G10 (FR4) was used on top of the PCB, mica, cardboard or other electrical insulation could be substituted. Each spring clip exerts a force of >60N.
The single clip mount is quite secure no extra mounting hardware is required to retain the PCB. The 3 switch DIP switches are in the mail so 4 way DIP switches were fitted for these tests.
The above image is an end view of a single clip mount onto a heatsink, often the enclosure is a convenient heatsink.
The image to the right is of the FET top mount option, good for all applications under 5 amps.
The image above shows the new PCB inside an old 33 mm vibrator case, the PCB has about 1.7mm distance from copper traces to the board edge ensuring it will not short to the metal case, top mounting of the FET's is recommended. The ambitious could design a thermal coupler for bottom mounted FET's enabling 10A continuous operation (240W @ 24V). When mounting in a vibrator case is is best not to fit the green terminal block and solder wires direct onto the board, see the image near the bottom of this page.
Above: a 200VA transformer under test, unfortunately my power supply maxes out at 6 amps and the transformer primary is 12-0-12 so it was not much of a test. The plan is to fit an over winding to produce an input voltage of 24-0-24 and parallel another power supply so it will run to 200VA. This transformer seemed better at 100Hz than 50Hz, The heatsink did not get warm at 70VA input.
I measured the DC resistance of the 12 volt winding of the 200 VA torroidal transformer at 57 milliohms, add to this the RDSon, fuse resistance and ESR of the supply capacitor to have some idea of the maximum current possible at turn on or fault condition.
This is a bottom view of the PCB one of the 2512 Vdd to PIC dropping resistors is shown at the bottom right, at the top of the PCB are 2512 pads for 1 watt snubber resistors and 1206 pads for snubber capacitors (untested).
The image to the right shows a PCB with top mounted FET’s driving a Jaycar transformer MM2015. The power supply is at maximum output with the current limit LED flickering at 6A.
The transformer was designed for sine wave operation at 50Hz, driving with a square wave introduces a lot of higher order harmonics to the mix. The 12V-01-2V taps are used as the primary and the 240V winding is output to a full wave bridge rectifier. The rating of the transformer is 100VA running like this < full power should be used about 75% is normally OK.
The high leakage inductance of the spit bobbin design is most suitable for a 50Hz sine wave and is not the best for 50Hz square wave. The DUT was run for a couple hours at 55 watts into the dummy load, an IR heat gun recorded a transformer temperature of about 50 degrees C, the FET’s were about the same temperature. No snubber components were fitted, the initial voltage spike at FET turn off was limited to 38 volts by FET avalanche characteristics, at full power the duration was 34 microseconds, the FET gate resistors are now 470 ohms to soften the switching slightly, cross conduction is not an issue due to the programmed off time 5%, 10%, 20% and 30% settable with the DIP switches.
I will endeavor to select FET’s rated for operation in the avalanche region, those skilled in the art could fit snubber components, pads are in place for two 2512 1W SMD resistors and two 1206 SMD capacitors. Running at 100Hz seemed to cause more heating of the laminations, all toroidal transformers tested seem better at 100Hz and could be run at a higher Vin than normal without core saturation. I have a created a spreadsheet for calculating core saturation.
Leakage inductance was not a problem with any torroidals tested, when using 60V FET's no avalanche operation. It should be noted vibrator transformers normally had low leakage inductance.
Running a 50Hz transformer at 60Hz places the core flux density closer to the original design value. Remember if running a 200V square wave into a bridge rectifier and capacitor the output voltage will be 200VDC, running a 200V sine wave into a bridge rectifier and capacitor the output voltage will be 282VDC.
New firmware written last night has these DIP switch selectable options:
50Hz at off times of 5%, 10%, 20% and 30%
60Hz at off times of 5%, 10%, 20% and 30%, 60Hz is normally the best choice for a 50Hz E I transformer.
For all continuous operation over 5 amps a heatsink is recommended. The piece of aluminium used for the dual clip mounting option should be enough for 10 amp operation, 10 amps is the PCB maximum, determined to some extent by the rating of the green terminal strip.
The image to the right shows how the PCB is to be wired. Most vintage radios have a fuse in the power supply wire, this is all that is required circuit resistances limit surge currents to safe values for the power FET’s, a typical DC resistance of the primary transformer winding is > 100 milliohms, a fuse is normally >10 milliohms.
If operating at higher power levels or with a high capacity electrolytic for C1 a fuse is recommended in the transformer centre tap. C1 should be bypassed with a non polar high frequency low ESR capacitor typically 1uF.
I measured the 12 volt winding of a 200 VA torroidal transformer at 57 milliohms DCR, and the inductance at 6 millihenries, this will be fine with a 10A fuse at the transformer centre tap.
Note: High current flows in the Q1, Q2 and GND terminals, <50mA flows in the +Vin terminal so the wire need not be very heavy, +Vin is current limited by resistors.
Leads from the transformer to the PCB should be as short as possible, remember the enclosed area rules :-) , C1's earth point should be the same as the GND terminals earth point.
Normally a voltage doubler would be employed on the output so a 12-0-12 volt to 240V transformer will supply a B+ of 450VDC when fed with ~12VDC.
30 Oct 2016
Many old electronic devices that were required to run from low input voltages employed a mechanical device that converted DC to AC enabling step up transformer to be used. The Vibrator made a buzzing/vibrating sound hence the name.
Below a typical circuit.
A typical 60Hz vibrator as used in a valve (radio).
This image is from Wikipedia.
https://en.wikipedia.org/wiki/Vibrator_(electronic)
My device could replace the vibrator in this circuit.
This vibrator schematic this version switches the DC input rail; my device would not be a direct replacement for this PCB.
Below is the schematic of my 2016 PCB.
As programed at the moment the DIP switch selections available are:
50Hz
60Hz
70Hz
100Hz
150Hz
200Hz
250Hz
300Hz
5VDC to 10VDC input; good gate drive at 5VDC input
7VDC to 14VDC input; good gate drive at 7VDC input
Note: The OFF time between driving each FET’s is set at approximately 300 microseconds.
This is an image of the device running a 240VAC lamp at a low level 7VDC input. The test lamp was ran at 40 watts for 3 hours and the PCB was about blood warm. 40 watts is the maximum recommended power. The light bulb is a compact 40W and is only 45 mm in diameter. Use a 3A fuse to the transformer center tap. The 9 volt battery is for scale. Note: 40W max 12 to 14V input only ( I^2R loss ).
The circuit board has no DIP switch installed as they have not arrived; I edited the program so without the DIP switch the frequency is at 60Hz, it is better to run high on frequency to keep the flux density down on the square waves.
If the output connector is not fitted the wires can be soldered directly to the PCB desirable in many cases. The programming connector is only required by those skilled in the art of microprocessor programming.
Software and tools used:
LTspice: Simulations.
DipTrace: Schematics and PCB design.
Iteadstudio (China): PCB production.
MikroC Pro: Firmware
Assembly: Myself
The mages below show the device fitted to a Ferris valve car radio, the B+ voltage is almost identical to the mechanical vibrator, the input current at 12.6 VDC has been reduced it was ~3A and is now ~2.5A.
No modifications were made to the radio, all the original snubber capacitors c/w associated components are in place. The frequency is set for 100 Hz as per the original vibrator.
The drain waveform looks good, device temperature is hard to judge as the entire radio warms up after a few hours, temperature is not a problem. The radio has been running for quite a few hours and the PCB is about the same temperature as the rest of the radio.
Some sort of metal case may be beneficial, the back side of the PCB is mostly one large ground plane and it seems very quiet electrically.
No programming header is required unless the user wants to reprogram the micro.
Set for 100 Hz with a jumper as the DIP switches have not arrived.
Ready for a plug
Ken's electronic Vibrator (upper right) running in a vintage tube car radio.
A size comparison between the old mechanical vibrator and Ken's solid state device.
A dog statue in my parents garden.
NOTE:
To my Australian friends, ten boards were produced for testing and some extras are available for beta testing, please give me a ring if interested.
The firmware is not in the public domain.
When using a mains frequency rated toroidal transformer on a square wave it can often be an advantage to increase the operating frequency. After doing the maths on flux density it can be seen one can increase the power rating of the transformer, the current rating will remain the the same however the voltage rating of a particular winding may be increased at a new higher operational frequency hence more power through the transformer.
When using this PCB a 6-0-6V to 240V 50 Hz transformer will normally be ok at 100 or 150 Hz @ 12VDC input for a B+ of 480VDC (full wave bridge).
BTW most vibrator power supply transformers have a low leakage inductance as do most toroidal transformers.
Ken Kranz
Adelaide
Australia
Disclaimer Although I will endeavour to keep all information accurate mistakes will be made. No claims of safety are made and all use of this information is at the risk of the user/builder. Please use the spice simulation and component data sheets to test component dissipation and other important parameters prior to building.
Page updated
Google Sites
Report abuse