Making the D5 Strong Variable speed

[Master Chicken]

Well, initial testing is done. This one makes 24.009 volts at the output with nothing plugged into the Fan Control header. Once loaded to 70 watts, it still makes the same with 9.6mV of ripple. The ripple decreases at lower loads. At that point, my input is at 11.7 volts. The inductor gets to about 96degC at the full 70 watt load. It's rated to 125degC. These tiny Carbonyl Iron inductors are amazing. This one is 6mm square and takes up 1/4 of the space that the prior board version used.

Once I try to crawl past 72 watts on the load, the IQDSM starts to dial back the voltage to maintain the operating point. This is the input current limit kicking in.

The Green LED is only run at about 1/2 normal power and I used a diffused lens so it's not very obtrusive. But it doesn't go out until the voltage is down by at least 9% from the target when the current limit is in effect. I suppose that'll have to do.

I probably need to dial back the snubber capacitors a bit. The switch node ring on the output side is lower in amplitude compared to the prior version and over-snubbing wastes power. I err on the side of bigger when designing so I don't destroy the FETs when first testing. Once I recalculate the snubbers, I'll be good to make the other test boards.

Tomorrow I'll hook up a PWM test tool and check the control characteristics and the ring on the input switch node at the lower output voltages (< 12V).

Eventually, of course, I'm going to have run a D5 Strong with it using the bucket of water test rig.

Now though ... I have to go watch football.

[Master Chicken]

I just finished reading the documentation. WOW, what Hooded said! Now you have me interested in turbocharging a D5 strong, or two!! Very well written documentation too. I just happen to have two of these things, and I'm wondering how much of a difference running em at the full 24v makes.

I'm just speculating here, but I'll bet they actually top out at something like 22.5 volts and that they don't use the full 24 volts. I won't know until we try to drive one. We'll be able to tell if the Tach frequency stops climbing at some point on the way up to 24 volts.

Or ... maybe fast_fate has already tested this and we just have to find a chart of his somewhere. I know he's tested a lot of pumps over the last few years.

[Master Chicken]

Did more testing at half time. The low voltage is 8.08 volts at 34% PWM. Close enough. My 5 volt rail on this ATX PSU is at 5.11 volts so it's a bit off. My 12 volt rail is at 11.83 volts lightly loaded.

The only thing not working right at the moment is the turn-off feature. Technically, it works but the hysteresis is 2% instead of 10% like it should be. I suspect a math error on my part. It should be correctable with a resistor value change. Of course, it's also possible that my kid put the wrong resistor in place ... so I've got to check that as well. So it works but it goes off at 20% and back on at 22% instead of the expected 10% and 20%.

[Bartacus]

I have digital flow meters in both my systems, so I'd be curious to see if running at 24v would result in faster flow rates, not that it would make a difference in terms of temps mind you.

[Izerous]

I like the idea that it can run two pumps (even if not fully maxed out). I had looked at picking up a pair of controllers in the past however I figured if in series getting them both running at identical speeds would be key, and would also be difficult with 2 separate knobs to get just right.

[Master Chicken]

I have digital flow meters in both my systems, so I'd be curious to see if running at 24v would result in faster flow rates, not that it would make a difference in terms of temps mind you.

Yeah, I think if you've got a big system including GPUs with waterblocks in SLI and 3 rads, CPU block, etc, all in one loop you might benefit a bit. Above a certain flow rate, going higher probably isn't beneficial once the loop settles but a higher rate is probably going to make the system more responsive to a step in the load. Even though it would settle at the same point, it will probably reach that point faster.

I'm planning on doing a TEC block in my system and I think a higher flow, when demanded, will help the step response of the system I'm planning. Plus, I get to build the Illudium Q-36 Thermo-Electric Modulator or IQTEM for that so ...

[Master Chicken]

I ordered more 0603 caps yesterday so I can futz with the snubbers. The smallest I have on hand is 2.2nF and I was hoping to land closer to 1.2nF. So I've got an assortment coming that should be here Wednesday.

[Bartacus]

Yeah, I think if you've got a big system including GPUs with waterblocks in SLI and 3 rads, CPU block, etc, all in one loop you might benefit a bit. Above a certain flow rate, going higher probably isn't beneficial once the loop settles but a higher rate is probably going to make the system more responsive to a step in the load. Even though it would settle at the same point, it will probably reach that point faster.

I'm planning on doing a TEC block in my system and I think a higher flow, when demanded, will help the step response of the system I'm planning. Plus, I get to build the Illudium Q-36 Thermo-Electric Modulator or IQTEM for that so ...

My system is a hot mess, but it has lots of stuff in the loop (4 rads from 3 different mfgs, 2 pairs of QDCs, 2 GPUs with a parallel block, CPU, 2 temp sensors, and the flow meter. With 2 D5 varios running at full blast, I get between 3.1 and 3.4LPM. It would stay constant at 3.4, but one of my varios has been flaky since install and it fluctuates at times, even when run by itself. If I run on a single pump, the flow drops to 1.7. I've got so much rad space for so little heat that it's hard to gauge the impact of 1 pump vs 2, pump speed, etc.