Help with Water Pump

[alec_t]

I gather it needs to be similar to mosfet IRF3205

I assumed you would be using IRF3205, as you have already sourced them. They tick all the boxes. The body diode is part of the IRF3205 (not separate as shown in the simulation schematic. So don't go hunting for body diodes.)

Does negative going edge mean the ground is briefly disconnected?

It just means the FET inside the pump switches off. We don't know whether that FET is a p-channel type (in which case it effectively disconnects the positive supply) or an n-channel type (in which case it effectively disconnects ground). However, for our purpose it doesn't matter; the switch off is just used as a signal to ensure M1 switches off for a set time.

 

[ronv]

1. Placement of D1 doesn't matter

Actually it should also be on the other side of D2 -- Right across the motor like every clamp diode you see.

2. Remove D1 and then use a UNIDIRECTIONAL TVS and the choice and direction of the TVS will matter.

Could do that. The TVS will allow the voltage to go higher than the clamp diodes however. We were trying to keep the voltage low across the motor??

 

[KeepItSimpleStupid]

The extra resistance would be just that extra resistance. The diodes can't be in the pump where they belong.

Your right on both counts. The schotkey (sp?) would have a much lower voltage drop or clamp in the neg direction, but the TVS is designed to handle surges a bit better.

 

[KeepItSimpleStupid]

Do you guys have a thermal design for M1? Heatsink size?

Joe: Are you aware of insulator kits for M1 (TO-220 packages) https://www.digikey.com/product-sea...hermal-accessories/1180797?k=to-220 mounting

and Thermal Grease: https://www.digikey.com/product-detail/en/860-150G/473-1097-ND/2233046
Radio Shack may have a tiny tube of grease. The stuff is almost impossible to get off your hands.

There are some thermal washers that don't require the grease: e.g. https://www.digikey.com/product-detail/en/SP2000-0.015-00-54/BER201-ND/529918

The "old school" way was to use Mica insulators. Better materials such as Kapton is now used and then some material doesn't require the grease.

 

[salty joe]

I assumed you would be using IRF3205, as you have already sourced them. They tick all the boxes. The body diode is part of the IRF3205 (not separate as shown in the simulation schematic. So don't go hunting for body diodes.).

I have a handful of IRF3205 in my parts box.

It just means the FET inside the pump switches off. We don't know whether that FET is a p-channel type (in which case it effectively disconnects the positive supply) or an n-channel type (in which case it effectively disconnects ground). However, for our purpose it doesn't matter; the switch off is just used as a signal to ensure M1 switches off for a set time.

Amazing!

Kiss-I was just going to take a heatsink from the OEMC and screw it to the IRF3205. Do I need grease for that?
Ronv, thanks for kicking this around.

 

[KeepItSimpleStupid]

I mentioned the body diode was internal eons ago.

Thermal grease should still be used. Radio Shack should have a small tube in the store: https://www.radioshack.com/product/index.jsp?productId=2102858 You use a very tiny amount. Just enough to fill the imperfections. If you don't have a latex glove, use a baggie or something. The stuff gets all over everything.

The tab is the drain, so it does put it in a position to get damaged. i.e. a short to ground.

Your choice though.
A thermal pad would electrically insulate the device from the heatsink.

If we modify the OEMC to get you an adjustable supply, you will need the heatsink there too. Again, your choice.

 

[alec_t]

Thermal grease should still be used.

IMHO special 'thermal grease' is a rip-off. The layer is so thin that almost any grease, or anything thin and electrically insulating that is soft enough to fill up surface imperfections will have negligible thermal resistance and will do the trick, e.g. Vaseline, ptfe tape.......

 

[KeepItSimpleStupid]

Anyway, **broken link removed**

which discusses toothpaste and vaselene.

 

[()blivion]

Off Topic, but related to recent posts
I agree 100% about there NOT being supper thermal paste that makes a difference. In fact, I know better from experience.
Someone, I don't remember who, pitted 20% percent zinc-oxide face cream up against the best of the best "magic" Thermal compounds with no measurable difference. As a PC technician by trade, I can tell you it's mostly just a marketing gimmick to make more money off the so called elitists that "know what they are doing". More important than the compound is that the two surfaces be flat and clean.

Water cooling a PC is also mostly snake oil. Think of this... what cools your water? OH WAIT! That's right..... the air you could have been using -$300 and 20lbs ago. Looks sweet though bro.

And finally, no one has ever made their PC a useful amount faster by making the CPU 0.1c cooler. Or even 1c for that matter.

(sorry for being off topic)

 

[KeepItSimpleStupid]

Big difference between zinc oxide paste and toothpaste.

Water cooling has it's place. Pop Mech did a nice CPU in a tool cabinet deal. Looked really slick. Water cooled. Oh, air to water heat exchanger. The PC didn't see the shop dust.

 

[salty joe]

Is this the arrangement of regulator 7812 and caps I should use to get down to 12V?

**broken link removed**

 

[()blivion]

Yes, keep in mind that C4-C5-C6 are "decoupling caps". They go on any chip, RIGHT on the V+ and Ground pins, as close to the chip as possible. This is done to isolate them from the noise on the power line. It "decouples" them from the power supply.

Edit: BTW, the 78 is the family of device, the 12 is the voltage.

 

[KeepItSimpleStupid]

The small caps are close to the 40xx series IC's. Between power and ground of those IC's.

 

[salty joe]

Thanks guys.
My little piezoes got here and a 9V battery makes a healthy beeeeep. They can take 12V. All the parts are ordered-can't wait for round two.

 

[salty joe]

Yes, keep in mind that C4-C5-C6 are "decoupling caps". They go on any chip, RIGHT on the V+ and Ground pins, as close to the chip as possible. This is done to isolate them from the noise on the power line. It "decouples" them from the power supply.

Edit: BTW, the 78 is the family of device, the 12 is the voltage.

The small caps are close to the 40xx series IC's. Between power and ground of those IC's.

Thanks for that heads up. I had been soldering the + leg of the decoupling caps directly to the IC's Vdd and soldering the - leg of the cap to ground. It sounds like I should try to get the - leg of the cap closer to the IC's Vss.

 

[alec_t]

Joe, if you're going for the post #812 circuit build then I suggest you make allowances to bring out two connections as 'hooks' for the future speed control; one (I call it 'SpeedCon') from the junction between R1 and R2, and one ('Isense') from the junction between R3 and Rsense.
I'll post a design for the speed control when I've explored a few more options. The current incarnation allows two speeds to be set; a 'fast' speed and a 'feed' speed. The 'fast' speed relies on a pulsed current with an average value about 80% of the continuous current which a direct 24V supply would drive through the pump motor. This has a similar effect to driving the motor continuously with ~18V (hence at a similar average power level to the OEM control, which both KISS and I think advisable for pump longevity), and eliminates the need for a further voltage regulator to give 18V (although for testing purposes in the meantime a simple mod of the OEM controller will give 18V).

 

[KeepItSimpleStupid]

I'll work on what mods you have to do to the speed controller to make it somewhat variable. I already gave you the list of the parts you needed.

On subject of "bypass caps" generally, the idea is to give the IC a little boost when it needs a little extra current to change states and to prevent the switching noise from propagating. Uncompensated OP amps may oscillate when the caps are omitted. The general reason that the cap is placed close to the IC is that adding lead length adds an infinitesimal amount of inductance to the circuit. In very high frequency circuits this matters.

In your circuit, you don't have to go through heroics to place the capacitors.

On another note, paralleling caps like a 100 uf electrolytic, 0.001 Ceramic, a 0.1 uf Tantalum and a even a 0.01 uf Polypro change how the combination behaves with frequency. This, again, is an esoteric application and a lot of times the values are empirically determined.

I can say without reservation, that caps between 1 and 10 uf and some audio amplifiers that should be metalized polyester or another type other than electrolytic, the electrolytic was used because of a lower cost. Vintage capacitors may also be physically bigger than today's counterparts.

 

[salty joe]

Thanks Alec, can you tell me about how many components I need to save room for? I could leave a spot to solder between those resistors and then leave enough room on the board for the additional components. Does that sound like a decent plan?

Thanks KISS, I could probably solder the - cap lead to the IC's Vss just as easily as taking it to ground. Does it matter?
Other than than the - lead from the pump, is it OK to run a continious 22 g wire ground loop for everything else? I have 19 g wire, should that be used for the pump leads and maybe the ground loop too?

 

[KeepItSimpleStupid]

Basically, follow the 3 AMP path current would flow through the motor. Those wires should be heavy gauge. The line drawn from From the TVS to the 0.22 resistor. The POWER GND on post 812 should be returned to the power supply negative terminal as a heavy wire.

The LOGIC GROUND from the other circuit should all be wired together. Again, returned to the power supply negative, but here the wire can be smaller.

Grounds, in general should be larger traces than most of the other circuitry, so yes the 10 AWG would be OK.

You can do POSITIVE in much the same way. The TOP of the TVS or the higher current goes directly to the POSITIVE terminal. The other POSITIVES which don't carry the motor current can be returned to the POSITIVE of the power supply.

For the bypass cap in this case, it really doesn't matter. Take whatever ground area is closer.

 

[alec_t]

can you tell me about how many components I need to save room for?

I envisaged all the speed control stuff as a separate module for handling 4 pumps. Since the design isn't finalised I'm not sure how many components will be used (but I've re-jigged the design to minimise components, hopefully), so I suggest you allow three 3 terminals of a multi-way terminal block for hooking into the speed module. Presumably you will have a terminal block for making all connections to/from the post #812 PDM?