Capacitor Charger LT3750

Thread Starter

MGPERF

Joined Jun 8, 2021
67
Hello,

I am looking to build a ~20 amp capacitor charger using the LT3750 however they do not specify any suitable transformers above 10 amp peak primary current. The LT3750 is rated for 10uH primary inductance. However the transformer I would like to use (coilcraft ga3459) is rated for 5uH. It does work in spice which is hopeful. Does anyone foresee any issues? In spice I keep reducing the value of R8 all the way down to 2m to get it to draw 10 amps. According to spice you’d never get anywhere near 10 amps with R8 at 8m per the reference design. Any harm in reducing the value that much?
 

Attachments

Last edited:

dcbingaman

Joined Jun 30, 2021
559
The circuit appears to be working correctly according to the datasheet:

1633049500346.png

Maximum current in amps being as shown 78mV/2m=39Amps peak. Which the simulation shows.

I changed it to 12m ohms for 6 Amps per datasheet and peaks are then 6Amps?

Not sure I understand what you are asking?
 

dcbingaman

Joined Jun 30, 2021
559
I am observing current peaks through the transformer primary as expected at 39A and being this is a 1:10 turns ratio observing 3.9A peaks on secondary.
 

dcbingaman

Joined Jun 30, 2021
559
Hello,

I am looking to build a ~20 amp capacitor charger using the LT3750 however they do not specify any suitable transformers above 10 amp peak primary current. The LT3750 is rated for 10uH primary inductance. However the transformer I would like to use (coilcraft ga3459) is rated for 5uH. It does work in spice which is hopeful. Does anyone foresee any issues? In spice I keep reducing the value of R8 all the way down to 2m to get it to draw 10 amps. According to spice you’d never get anywhere near 10 amps with R8 at 8m per the reference design. Any harm in reducing the value that much?
To reduce risk of it being damaged, I would stick to the transformers recommended per the LT3750 datasheet.
 

dcbingaman

Joined Jun 30, 2021
559
If you want 20A peak into the output capacitor you will need 200A peak on the primary side. That is way to much current for most transformers. For the ones recommended for the part the max primary current is 10A so the max secondary current will be only 1A. You are going to need a different design if you want to meet your 20A charge requirement.
 

Thread Starter

MGPERF

Joined Jun 8, 2021
67
This could work then, I am going to be putting 15 volts in and would need about 150 watts of power max for brief periods of time. Per the attached reference sheet when they call it a 9 amp capacitor charger what exactly does the 9 amps refer to? If the transformer primary is rated for 10 such as in this reference design then this could only draw something like 5 amps right?
 

Attachments

Thread Starter

MGPERF

Joined Jun 8, 2021
67
Also if I am putting 15 volts in and getting 300 plus volts out with a 1:10 turns ratio is there something that steps the voltage up past 15 at the transformer primary? It's not showing it in spice however that would help me reach my power output.
 

ronsimpson

Joined Oct 7, 2019
2,014
Look at the 3, 6 or 9A examples. Picture below is from the 9A example. Clearly the current is 9A peak. That is about 2.5A from the 12V source so 30 watts. Or at 300V that is 0.1A.
1633059540319.png
 

ronsimpson

Joined Oct 7, 2019
2,014
Also if I am putting 15 volts in and getting 300 plus volts out with a 1:10 turns ratio is there something that steps the voltage up past 15 at the transformer primary? It's not showing it in spice however that would help me reach my power output.
From your spice file. Green trace is the voltage on the MOSFET. The voltage is "flying back" to 47V. So the primary sees 15V when the MOSFET is on and 47-15=32V when the MOSFET is off. 32V x 10 =320V.

That math is not quire right but close. You are loosing power in the HOT mosfet because you picked a too mall a part for 60A and there is loss in the diode and resistance in the transformer.
1633059874514.png
 

Thread Starter

MGPERF

Joined Jun 8, 2021
67
Looking like I’ll actually have to go with a Coilcraft ga3460 which has even lower inductance to get the power output I need. I am going to look into increasing resistor values for Rout, RDCM, and source. I’m going to use the diodes and mosfet from the 42 amp capacitor (lt3751) capacitor charger reference design.
 

dcbingaman

Joined Jun 30, 2021
559
Looking like I’ll actually have to go with a Coilcraft ga3460 which has even lower inductance to get the power output I need. I am going to look into increasing resistor values for Rout, RDCM, and source. I’m going to use the diodes and mosfet from the 42 amp capacitor (lt3751) capacitor charger reference design.
The LT3751 looks like an excellent part to use with the recommended 42A charge circuit in datasheet. I noticed they have a recommended layout for the PCB. Naturally with a circuit with so much current and power layout is critical. Sounds like you are implementing the circuit on page 25 of the datasheet with some modifications to the feedback sense resistor to achieve 20A? Being the GA3460 is one of the recommended fly-back transformers, should work fine as long as PCB layout is well considered. Much success.
 

Thread Starter

MGPERF

Joined Jun 8, 2021
67
I am pretty sold on using the LT3750 because it’s so much easier to lay out (less pins) and it’s ok to go over 300 volts so long as you put in a larger resistor for RDCM. Just curious about putting much more current to it vs. recommended however I am reducing current appropriately by putting in larger resistors.
 

dcbingaman

Joined Jun 30, 2021
559
Also if I am putting 15 volts in and getting 300 plus volts out with a 1:10 turns ratio is there something that steps the voltage up past 15 at the transformer primary? It's not showing it in spice however that would help me reach my power output.
Because we are dealing with square waves with the MOSFET turning off the circuit is designed to convert primary voltage up due to V=L*dI/dt or in this case looking at the change in current in LTspice from graph it looks around 30A/10ns or 3e9 Amps per second then V=2.5uH*3e9 = 7,500 Volts. Naturally the transformer energy is being thrown into the capacitor via the transformer it never reaches this high voltage limit but will reach voltages higher than the input voltage. The chip controls the limit and after the cap is charged stop operation as expected.
 

dcbingaman

Joined Jun 30, 2021
559
Ran an interesting experiment. Basically removed the output capacitor. The chip must try one cycle to find out what the voltage is during charging. Being it has no idea there is no energy transfer on the secondary the voltage on primary jumps up to 440V! Still not even close to the theoretical 7,500V. Probably the MOSFET capacitance is keeping it from reaching the theoretical 7.5KV.
Looked at the di/dt again. With no output capacitance and reducing the transformer capacitance to 0, it appears around 30A per 300ns or 100e6 volts per second for a primary voltage of 2.5uH*100e6 = 250V. I am estimating it here by 'eyeing' the graph so I am off a little with it actually reaching 440V. It appears the maximum slope is around twice what I calculated.
 
Last edited:

dcbingaman

Joined Jun 30, 2021
559
I am pretty sold on using the LT3750 because it’s so much easier to lay out (less pins) and it’s ok to go over 300 volts so long as you put in a larger resistor for RDCM. Just curious about putting much more current to it vs. recommended however I am reducing current appropriately by putting in larger resistors.
I am fairly certain the layout is designed per the datasheet maximum current of 10A on the primary. Even the datasheet proposed layout for the LT3750 could run into problems if you try to push it up to 20A which is probably why the LT3751 has a different layout to handle higher currents. I would recommend if using the LT3750 to simply set the primary at maximum 10A. I know you wanted 20A but I don't think it is achievable with that part. Also I would not use the LT3751 recommended 2.5uH transformer with the LT3750. Sure it may 'look' like it works in sim, but could damage parts because it is not recommended on the datasheet. The LT3751 allows lower inductances to 2.5uH but the LT3750 allows a minimum inductance for proper operation of 10uH on the primary.
You are correct to be concerned about to much current. It exceeds datasheet recommendations. I would not do it. If this circuit requires high reliability over the long run there is no guarantees after you step outside the boundaries of the datasheet.
 

dcbingaman

Joined Jun 30, 2021
559
You can build two supplies and run them in parallel.
Great idea. It works out perfectly being the diodes only allow sourcing no sinking, would work great and would meet all the specifications of datasheet for LT3750 placing the two 9Amp circuits shown in the datasheet in parallel.
 
Top