My summary results from back then were:
0-110-220V full-wave with CT rectified output:
10V 185mA 1.83W; 99V 560k (18mW); LED on, no load
14V 263mA 3.7W; 144V, no load
UVLO at 7.5Vdc
0-280V full bridge:
10V 173mA 1.73W; 248V 1M (62mW); LED on, no load
14V 239mA 3.4W; 335V no load
14V 452mA 6.3W; 325V 34k 9.6mA 3.1W load
14V 581mA 8.1W; 323V 20.4k 16mA 5.1W load
0-220V doubler rectified output:
10V 190mA 1.9W; 397V 2Megohm (78mW); LED on, no load
14V 273mA 3.8W; 548V, no load
12V 686mA 8.23W; 443V 34k 13mA 5.78W load ; 2.45W loss
12.19V 730mA 8.9W; 448V 34k 13mA 5.9W load ; 3W loss; 0.022 ohm sense; mp3242 powertech plugpack
12.0V 1740mA 20.9W; 441V 10.6k 42mA 18.5W load ; 2.4W loss
11.8V 3.77A 44.5W 53mVrms; 426V 4.4k 97mA 41.2W load ; 3.3W loss; fet hs temp rise abt 3-4C. tx winding temp rise abt 20C.
Drain waveform transition time was about 1uS.
I'm not sure about the 'saturation' description. Idle losses certainly increase from 12V to 14V. A small part of that is due to primary and secondary side capacitance charging/discharging. Part of that will be the higher final peak primary current (assuming constant inductance) causing a increased loss in FET and primary winding and core loss. It may be quite difficult to determine if the FET current slope starts to deviate from a linear ramp at a higher input voltage.
From a pcb layout perspective, there are a few convenient places for smt caps.
0-110-220V full-wave with CT rectified output:
10V 185mA 1.83W; 99V 560k (18mW); LED on, no load
14V 263mA 3.7W; 144V, no load
UVLO at 7.5Vdc
0-280V full bridge:
10V 173mA 1.73W; 248V 1M (62mW); LED on, no load
14V 239mA 3.4W; 335V no load
14V 452mA 6.3W; 325V 34k 9.6mA 3.1W load
14V 581mA 8.1W; 323V 20.4k 16mA 5.1W load
0-220V doubler rectified output:
10V 190mA 1.9W; 397V 2Megohm (78mW); LED on, no load
14V 273mA 3.8W; 548V, no load
12V 686mA 8.23W; 443V 34k 13mA 5.78W load ; 2.45W loss
12.19V 730mA 8.9W; 448V 34k 13mA 5.9W load ; 3W loss; 0.022 ohm sense; mp3242 powertech plugpack
12.0V 1740mA 20.9W; 441V 10.6k 42mA 18.5W load ; 2.4W loss
11.8V 3.77A 44.5W 53mVrms; 426V 4.4k 97mA 41.2W load ; 3.3W loss; fet hs temp rise abt 3-4C. tx winding temp rise abt 20C.
Drain waveform transition time was about 1uS.
I'm not sure about the 'saturation' description. Idle losses certainly increase from 12V to 14V. A small part of that is due to primary and secondary side capacitance charging/discharging. Part of that will be the higher final peak primary current (assuming constant inductance) causing a increased loss in FET and primary winding and core loss. It may be quite difficult to determine if the FET current slope starts to deviate from a linear ramp at a higher input voltage.
From a pcb layout perspective, there are a few convenient places for smt caps.
Have you measured the switching frequency? Ther are probably 2 different types made. If it is 37KHz as in the description in the link I submitted it is already modified and OK, if 25KHz it is not performing good and needs attention.
Also watch out the zobel between the primaries on the board between the MOSFET’s and the transformer, 10nF in series with 5.1 Ohm. The brown mylar cap was open circuit on my board.....
My board now consumes 100mA at 12.6V and 120mA at 14.5V in idling without zobel and load. Before at 25KHz and 12.6V it consumed 250mA...and 500mA at 14.5V
I’m still working on the zobels.
Johan
Also watch out the zobel between the primaries on the board between the MOSFET’s and the transformer, 10nF in series with 5.1 Ohm. The brown mylar cap was open circuit on my board.....
My board now consumes 100mA at 12.6V and 120mA at 14.5V in idling without zobel and load. Before at 25KHz and 12.6V it consumed 250mA...and 500mA at 14.5V
I’m still working on the zobels.
Johan
Last edited:
Yes I have early version units with 25kHz, as per post #1 details, and RT=4k3 and CT=5N6. They probably use 4n7 or 3n9 now.
My units have 22 ohm deadband (pins 5-7).
The 5R1 gate drive resistors are likely a good choice given the large loop path. A smt cap added across pins 12-13 is easy to add.
It will be interesting to see which parts get close to temperature limits in a practical implementation (eg. where the FET heatsinking is extended to a metal wall) - some assumptions would need to be made about the allowed max core temp, as the core material will be an educated guess.
My units have 22 ohm deadband (pins 5-7).
The 5R1 gate drive resistors are likely a good choice given the large loop path. A smt cap added across pins 12-13 is easy to add.
It will be interesting to see which parts get close to temperature limits in a practical implementation (eg. where the FET heatsinking is extended to a metal wall) - some assumptions would need to be made about the allowed max core temp, as the core material will be an educated guess.
Strange, both my units have 200 Ohm (red-black-black-black-brown) deadband resistors in original! I even measured them to be 199 Ohm as I didn’t belive it should be that high. I’ve now changed them to 27 Ohm.
You can also see the 200 Ohm beside the 4.3 KOhm on many photos in eBay/Ali........so it is not only in my units!
About the zobel capacitor (the small brown mylar 10nF) I think burnt off, as it could not carry the peak currents in the 5.1 Ohm 1W series resistor!
Still working with that....
Johan
Now I have decided the final solution how to modify my inverters.
I put back the 5.6nF timing capacitor, changed the 4.3KOhm timing resistor to 2.2KOhm instead to run the oscillator at 104KHz
Deadtime resistor changed from 200 to 27 Ohm.
Took away the original zobel network 10nF & 5.1 Ohm between the primaries of the transformer.
Put in 2 zobels, a 10nF ceramic discs in series with 8.2 Ohm 1/4W. One over each primary winding on the transformer.
The inverter now runs very cool at 52KHz switching frequency with minimal overshoot and no ringing on the squarewave.
Some measurments at konstant input battery voltage 12.8VDC and the on board full wave rectifier on the 220V tap and with a Sanyo 22uF/400V output capacitor.
No load total current consumption: 110mA @12.8VDC
3.14W load. 13mA. @ 238V. 340mA @ 12.8VDC
6.24W load. 26mA. @ 237V. 580mA @ 12.8VDC
9.42W load. 40mA. @ 237V. 820mA @ 12.8VDC
12.6W load. 53mA. @ 237V. 1050mA @ 12.8VDC
So the effiency at 12.5W load is already up to 93% and the total losses less than 1W at that power level. I assume no problems to load it to 50W + without extra cooling.
Johan
I put back the 5.6nF timing capacitor, changed the 4.3KOhm timing resistor to 2.2KOhm instead to run the oscillator at 104KHz
Deadtime resistor changed from 200 to 27 Ohm.
Took away the original zobel network 10nF & 5.1 Ohm between the primaries of the transformer.
Put in 2 zobels, a 10nF ceramic discs in series with 8.2 Ohm 1/4W. One over each primary winding on the transformer.
The inverter now runs very cool at 52KHz switching frequency with minimal overshoot and no ringing on the squarewave.
Some measurments at konstant input battery voltage 12.8VDC and the on board full wave rectifier on the 220V tap and with a Sanyo 22uF/400V output capacitor.
No load total current consumption: 110mA @12.8VDC
3.14W load. 13mA. @ 238V. 340mA @ 12.8VDC
6.24W load. 26mA. @ 237V. 580mA @ 12.8VDC
9.42W load. 40mA. @ 237V. 820mA @ 12.8VDC
12.6W load. 53mA. @ 237V. 1050mA @ 12.8VDC
So the effiency at 12.5W load is already up to 93% and the total losses less than 1W at that power level. I assume no problems to load it to 50W + without extra cooling.
Johan
Last edited:
Noob alert I'm afraid..
I've been tinkering with the RECOM units until now as I'm only making a 1W amplifier but was having cut out issues with the output...
If I'd like to try this unit "as shipped" will I get usable results before making the suggested revisions here or is that not advised?
I am building without a fancy scope at the moment so all measurments will be V, mA, and mk1 ear
I've been tinkering with the RECOM units until now as I'm only making a 1W amplifier but was having cut out issues with the output...
If I'd like to try this unit "as shipped" will I get usable results before making the suggested revisions here or is that not advised?
I am building without a fancy scope at the moment so all measurments will be V, mA, and mk1 ear
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.