Need help with a 24V to 48V boost converter.

[jackatfsi]

Ok...looks like I'm spending waaay too much time on this but I can't help myself......

This is a (still a little simple-minded) workup of the boost input-current controller using the 400W inverter dc-dc portion.

The input is limited to 12V, 50 Amps for this exercise and the transformer secondary is (to be) rewound for 48V with 36V and 24V taps. For the moment, rectification is by diodes with a .8V forward drop and the only resistance used is that of the load (ie .6 ohms for the Heinsmann 500W hub motor. RPM is output from the geared motor shaft.

The Voltage curves aren't very surprising, the power out looks like it would help with acceleration, but not as much as I'd like, the efficiency, though, looks really good at low throttle settings.........

These steps are probably more extreme than one might expect w/ a real-world version but they do show some promise......that is to say...., if the FET switches in the AC section could be rewired to be a synchronous PWM and follow the "optimum" Voltage/Current ratio then the those peaks might be joined in a continuous curve.......

For now it's back to the hardware to see how much of this is do-able.....

 

[jackatfsi]

Here are the pictures of the 12VDC (could just as easily be 24VDC) to 120VAC inverter..before I take it apart and build it into the up-converting controller...

The black ink is my not too successful attempt to trace out more of the circuit........

The plan at the moment is to start w/ a new proto board and remove and relocate the components as we go along. The DC-DC part will be very similar to what is there now. The transformer will be rewound to peak out at 48V and may have lower voltage (higher current) taps depending on the available room on the core.

The two inputs to the driver chip can be set up for input voltage cutoff (10.5V) and input current control (and limit). The transformer is to be wound on the basis of current output.

An interesting question at the moment is whether to add a series inductor to the output circuit........that is, make the output circuit act like a down-converter under high loads........(the current/power control is to be on the input side)

 

[fechter]

An interesting question at the moment is whether to add a series inductor to the output circuit........that is, make the output circuit act like a down-converter under high loads........(the current/power control is to be on the input side)

Hmm... I dont' think I've seen that topology. Usually they just let the transformer do everything. An extra inductor will introduce more resistance loss.

Those transformer cores are very brittle. Hopefully it is split so you can take out the windings. Some heat will weaken the glue. Usually there's not too many turns, so rewinding should be fairly easy.

What is the target input / output voltage?

 

[jackatfsi]

Well, the target input voltage is 12VDC w/ current ( = power) limit/control on the input. (this same transformer primary will work w/ 24VDC but would need 2 more FETs for an H setup)

I'm thinking of using a rewound and tapped secondary w/ turns ratios 1:1, 1:2, 1:3, 1:4 w/ decreasing wire size and chopping the output w/ the 4 output FETs to follow the impedance of the motor at different speeds.

This could be done w/o a transformer as a LC circuit but I think the transformer will be easier to try things w/ as I can make isolated circuits and separate boards initially.

 

[fechter]

Why not just control everything from the primary side? The output could just be a bridge rectifier and filter capacitor.

 

[jackatfsi]

That's ok, I want to control the power input from the primary side.....the output current possible will depend on the transformer turns ratio however........so as the speed varies, I need to be able to optimize the current to the motor.......simple switches can give me something like the last curve above, but modulating the taps may give me a smooth curve w/ the "best possible" match......thats where a series inductor "might" make things just a bit better.....(and the FETs lose less power than the diodes..in theory anyway)

 

[fechter]

I see. Does it use synchronous rectification? FETs will have much less voltage drop than diodes.

 

[jackatfsi]

Yes..that would be the best bet....esp since we have the 20KHz frequency generated on the primary and don't need to have 60Hz at the output

 

[jackatfsi]

Slow going w/ everything else happening around here.....but here's a shot of the proto board to be mounted in the heat sink...next step will be to transfer the primary side components set up as input current (ie. power)control....I think the fuse is going to double as the current shunt....we'll see

 

[dermot]

Have a look at the boost circuit by Dan N8XJK, a friend of mine in the UP, Michigan.

It is designed to handle 150W and supply a variable boost voltage to a 12V SLA to provide a regulated 13.8V for ham radio use. Since it only has to provide the 'top up' voltage by adding on the transformer-coupled output to the input voltage, 150W goes quite along way. I've been thinking about using a modified version of this to eliminate voltage sag from a conventional battery pack.

My NiMH 36v pack only sags by a handful of volts for most of a cycle, so one of these boards could easily provide, say, 3V at 50A to eliminate that much sag.

I've included the info here as a zip file - Dan gave me one of the PCBs he's designed together with the transformer. They are freely available to purchase.

The design is pretty straightforward to understand and, I think, to modify. A bigger transformer and FETs should get you where you want to go.

There is going to be a significant loss in the pair of diodes on the output, it may be that replacing the diodes with FETs and active commutation is the way to go.

 

[eP]

...
My NiMH 36v pack only sags by a handful of volts for most of a cycle, so one of these boards could easily provide, say, 3V at 50A to eliminate that much sag.
...

Are You kidding man ?

at 50A and 36V you need P_out = 1800 W.

Tour step-up DC-DC converter will suck 60A at least so the battery sag will rise to 3.6V.

Now you have 3V*50A =150W battery loss, after conwersion you get 5V*60A =300W of loses.
3.6V*60A=216W at battery + 1.4V*60A =84W at converter (at the best case ).

So you get TWICE more loses this way.

If you want really decrease battery sag and battery loss the best way is to add the next parallel pack.

 

[jackatfsi]

I'm still thinking about the "boost" configuration.....the big advantage is that the transformer limited supply only has to supply the boost power (the input FETS have plenty of capacity).......of course I'm looking at a 36V peak boost.....but it would eliminate the highest current part of the secondary winding

Synchronous rectification is a must if I want the output impedance to follow the motor load and speed (and avoid loss)........the part that still has me worried is the isolation and how to power the active devices on the output side (diodes make it no problem in the simple boost circuit)......still fussing over this

I'll go ahead w/ the input (total power) control anyway....the input will control the power and the secondary will match the motor between 12 and 48VDC one way or another (but it will be more efficient w/ the boost version)

 

[dermot]

...
My NiMH 36v pack only sags by a handful of volts for most of a cycle, so one of these boards could easily provide, say, 3V at 50A to eliminate that much sag.
...

Are You kidding man ?

at 50A and 36V you need P_out = 1800 W.

Tour step-up DC-DC converter will suck 60A at least so the battery sag will rise to 3.6V.

Now you have 3V*50A =150W battery loss, after conwersion you get 5V*60A =300W of loses.
3.6V*60A=216W at battery + 1.4V*60A =84W at converter (at the best case ).

So you get TWICE more loses this way.

If you want really decrease battery sag and battery loss the best way is to add the next parallel pack.

No! You don't understand the content of my post, where I used the word 'BOOST'.

All this circuit does is ADD a few volts on top of the main pack, so 150W will supply 3V at 50A.

ADD that ON TOP of the original pack voltage.

The boost idea is what makes the idea workable - and, in fact, very efficient since the electronics only handle the extra power required, not the whole supply. The output diodes, obviously have to handle the total output current, so need to be pretty beefy...

 

[dermot]

Synchronous rectification

*That's* the phrase I was grasping for early this morning....

I don't think this is a problem, think of a tiny DC-DC converter for power and an opto-isolator for switching teh FETs.

I will admit, however, that at this stage, the complexity begins to increase rather a lot.....

 

[jackatfsi]

Good point...could just be a small extra winding on the secondary w/ diodes to power the switching FETs...........I wonder if that might be the reason for the "unused winding" in the original configuration ?????

For that matter...might just couple the switching freq right off that winding as well.......

 

[eP]

No! You don't understand the content of my post, where I used the word 'BOOST'.

All this circuit does is ADD a few volts on top of the main pack, so 150W will supply 3V at 50A.

ADD that ON TOP of the original pack voltage.

The boost idea is what makes the idea workable - and, in fact, very efficient since the electronics only handle the extra power required, not the whole supply. The output diodes, obviously have to handle the total output current, so need to be pretty beefy...

Ok. It is better idea than that i think before.
But is 3V really worth of 150W DC-DC converter ?

The same you can gain if you add 3.7 V parallel Li-ion pack.

 

[dermot]

Ok. It is better idea than that i think before.
But is 3V really worth of 150W DC-DC converter ?

The same you can gain if you add 3.7 V parallel Li-ion pack.

(Firstly, I assume you meant serial to add 3.7V, not parallel?
Putting a single 3.7V Li-Ion cell in parallel with a 36V or 48V high-capacity pack would surely cause it to explode!!!)

Your suggestion would sort-of work for a quick and dirty solution (except that it would be unregulated, we'd still see a drop off in performance as the batts discharged - a boost-regulated solution would guarantee an absolutely constant 36.00V - or whatever, until the main pack was way down).

But I'm really only interested in long term, reliable and easy to maintain / charge solutions.

Needing two different types of charger for two different chemistry batteries would be a real pain - plus the all-too-real pain of the Li-ion pack exploding under my backside if it went flat before the the main pack and 30A reverse-charged the suicide bomb - sorry, Li-ion pack...

I think there's whole world of difference between what I might do for a bit of fun one summer's afternoon and what I would do for a long term modification to my main method of commuting!

It would be mandatory to have a DrainBrain (oops! charge-monitor-previously-known-as...) as you'd have no indication of state of charge, since the voltage would tend to stay very stable until the end.

 

[dermot]

Good point...could just be a small extra winding on the secondary w/ diodes to power the switching FETs...........I wonder if that might be the reason for the "unused winding" in the original configuration ?????

For that matter...might just couple the switching freq right off that winding as well.......

You're right, of course, a pair of overwinds, one for each phase, is a much more elegant solution!

One neat way of managing the overall design would be to have a pair of diodes as per the simple implementation, but with FETs in parallel.

The boost circuit is then arranged to normally be in standby, thus drawing no current. As soon as the motor draws current (through the diodes, from the 'unboosted' supply) a current sensor on the input side turns on the boost curcuit, but only as needed.

Without this, you'd need to keep manually switching the boost circuit on and off or the standby drain would kill the battery life.

 

[jackatfsi]

Hmmmm...I really like the diode switched idea....but I think I might want to turn it around......what I really want the boost to do is make up for the motor back EMF as the speed increases.....so I could measure on one half-cycle and pick one of 5 boost levels (0 thru 4) to use on the next half-cycle.......assuming I can set that up..presumably w/ zeners and an inhibit ladder so that only the highest FET is on (or none)