Low flow rate in loop with 4x MCP35x ?

[lms060]

I have a custom loop on my main pc with 2x480 rads 2xgpu blocks and 1 monoblock and with 2x dual swiftech mpc35x pumps and I'm getting less than 0.5 GPM on my loop at 100% on all pumps.

So would I be looking at too many bends or some blockage in one of components ?

Any suggestions are welcome smile.gif

 

[Flat-6]

Your loop is relatively simple and in no way does it have too many bends etc. Check out my DejaVu build (link in my signature below) - When I was testing DejaVu I test run an MCP35X2 and that achieved over 1.5GPM in that build...!

How are you PWM driving the pumps - I assume via an Aquaero?

Have you entered the calibration value for your flow meter into the Aquaero?

Need a bit more info before you assume some sort of blockage etc.

Just one of those Swiftech MCP35X2 pump blocks would be way more than adequate for your loop (These pumps combined with the Swiftech top are very powerful) - you definitely don't need two of them. Remove one and keep it aside as a spare.

Keep in mind that DDC pumps require some airflow underneath their housings - so ensure that you have some reasonable airflow underneath the pumps - Don't close off the bottom of the pump housings with too much foam etc.

 

[lms060]

Your loop is relatively simple and in no way does it have too many bends etc. Check out my DejaVu build (link in my signature below) - When I was testing DejaVu I test run an MCP35X2 and that achieved over 1.5GPM in that build...!

How are you PWM driving the pumps - I assume via an Aquaero?

Have you entered the calibration value for your flow meter into the Aquaero?

Need a bit more info before you assume some sort of blockage etc.

Just one of those Swiftech MCP35X2 pump blocks would be way more than adequate for your loop - you definitely don't need two of them. Remove one and keep it aside as a spare.

Keep in mind that DDC pumps require some airflow underneath their housings - so ensure that you have some reasonable airflow underneath the pumps - Don't close off the bottom of the pump housings with too much foam etc.

Loop being simple is the main thing that bothers me the most. The previous version of this build was with a lot more bends and custom reses and more components and with a EK-XTOP Revo Dual D5 I was getting over 1.6 GPM. The idea of this build was for it to be simple.

Calibration value is supposed to be the correct on for the Aquacomputer sensor I use.

As for rads I'm using 2x EK-CoolStream XE 480. Monoblock is the EKWB r5e10 one. And everything is with either 10 ID acrylic/soft tubing.

The only thing I can think of that could be really wrong is having some kind of blockage in one of the blocks/rads but I really don't have the time to take everything apart and clean all of them in a proper way since I use this PC as a work PC.

As for the airflow for the pumps the bottom push/pull blows at them and there is a front fan blowing at them as well. The foam under them covers only the center since I used single peaces for both pump. Not sure if thats ok or not.

 

[bigtallanddopey]

I used to work with hydraulic systems and I would guess the same theories apply to water cooling.

We would never use two pumps on the same line. Increasing the number of pumps doesn't increase flow rate at all. It may even be possible that they are working against each other as you will never match the speed of them both, there will always be a slight difference in speed.

Again this is from my hydraulic days but the same should apply.

Flow rate is determined by the size of the pipes, connectors and any components along the way. Determining how water flow reacts with all the components is a very tricky bit of engineering.

I would suggest running with one pump, take the other one out. Or split the loops. One gpu loop and one cpu loop.

Also as a side note. Are you sure that those gpu blocks are piped up right? Wouldn't you only require one pipe between the two? I've never water cooled an sli set up though so ignore this if it is the way it should be.

 

[Flat-6]

Increasing the number of pumps doesn't increase flow rate at all.

This is not correct as adding a second pump increases the pressure which for a given restriction will increase the flow rate.

There has been a fair amount of work done over the years wrt PC cooling loops and running two pumps in series is a well known configuration.

This is why you have dual pump tops such as the MCP35X2. Of course the law of diminishing returns apply so you cannot simply keep doubling the number of pumps to keep doubling the output pressure.

 

[bigtallanddopey]

This is not correct as adding a second pump increases the pressure which for a given restriction will increase the flow rate.

There has been a fair amount of work done over the years wrt PC cooling loops and running two pumps in series is a well known configuration.

This is why you have dual pump tops such as the MCP35X2. Of course the law of diminishing returns apply so you cannot simply keep doubling the number of pumps to keep doubling the output pressure.

You are right they increase pressure but not flow rate. That is governed by restrictions through the system. Larger pipes for example will produce a lower flow rate than smaller diameter pipes. It's a very difficult situation and lots to understand.

Two pumps in series shouldn't increase much though as a pump can only output what it can output it regardless of what it's input pressure is. This will be more true the more well made the pump is. To increase the amount of fluid coming out of the pump you either increase the pump speed or increase the size of the pump. Not just add another pump on.

 

[trents]

Another option would be to divide the system into two loops and utilize each pump for it's own loop.

 

[Flat-6]

You are right they increase pressure but not flow rate. That is governed by restrictions through the system.

Not quite - If you increase the pressure for a given restriction the flow rate will increase in our loops - If this was not the case then our flow rates would always be fixed no matter what pump speeds we select.

It is no different when we adjust the speed of our pumps - we can reduce or increase our flow rates.

Running two pumps in series such as the Swiftech pump top provides is a great way to increase flow rates and makes reasonable gains. See here: https://martinsliquidlab.wordpress.com/2011/04/26/pump-setup-series-vs-parallel/

It is not a linear relationship (ie. 2x pumps does not equal 2x flow rate) however the gains are still useful for our typical PC loops.

Here is Martin's results when he tested the MCP35X2 pump housing:




Full review can be found here: https://martinsliquidlab.wordpress.com/2012/01/29/swiftech-mcp35x2-pump/

 

[lms060]

So i decided to split the loops in 2 and the bottom 2xGPUs + 1 480 rad are getting ~1.6 GPM but the top loop of 1xCPU monoblock + 480 rad is getting less than 0.80 GPM.
I tried running the monoblock only without the rad and there is no increase in flow so it does seem the monoblock is creating almost all the restriction in the loop.

 

[trents]

So i decided to split the loops in 2 and the bottom 2xGPUs + 1 480 rad are getting ~1.6 GPM but the top loop of 1xCPU monoblock + 480 rad is getting less than 0.80 GPM.
I tried running the monoblock only without the rad and there is no increase in flow so it does seem the monoblock is creating almost all the restriction in the loop.

Some good troubleshooting there. Are you sure you have the intake and return hoses connected to the correct sides of the water block? Can the water block be disassembled and inspected?

 

[lms060]

Some good troubleshooting there. Are you sure you have the intake and return hoses connected to the correct sides of the water block? Can the water block be disassembled and inspected?

I think so, I've looked at the manual over 10 times during first install since it's kinda weird how the right hole is the intake.

 

[bigtallanddopey]

Can you test the flow rate just going through the rad and not the block? Maybe the pump just isn't working. Or maybe there is a blockage in the block restricting flow?

 

[lms060]

Can you test the flow rate just going through the rad and not the block? Maybe the pump just isn't working. Or maybe there is a blockage in the block restricting flow?

As I mentioned before I tested running only the monoblock and the flow didn't change compared to running monoblock + rad

 

[bigtallanddopey]

As I mentioned before I tested running only the monoblock and the flow didn't change compared to running monoblock + rad

I meant just the radiator not the block. A test for the pump to see if it is outputting what you would expect compared to the other pump.

 

[lms060]

I meant just the radiator not the block. A test for the pump to see if it is outputting what you would expect compared to the other pump.

Oh didn't get the point of the test before. will definetely test, just need to buy distilled water again (seems kinda hard to find in my area )

 

[maxfly]

make sure neither of your fittings are bottoming out in your monoblock. i had the same issue with really bad flow and found that both of my gpu blocks had fittings that were bottomed out and didnt allow much flow. replaced them and boom flow rate was right where it should be.

 

[lms060]

Can you test the flow rate just going through the rad and not the block? Maybe the pump just isn't working. Or maybe there is a blockage in the block restricting flow?

tested rad only in the current setup (just bypassed the monoblock) and it's doing 2.75 GPM

make sure neither of your fittings are bottoming out in your monoblock. i had the same issue with really bad flow and found that both of my gpu blocks had fittings that were bottomed out and didnt allow much flow. replaced them and boom flow rate was right where it should be.

Checked that as well and it doesn't seem to be the case.

 

[lms060]

I feel kinda dumb decided to take apart the monoblock, there was some gunk on it, cleaned it and now I'm getting 1.95 GPM.
Running only rad its 2.75, so by adding 30 cm of tubing and 4x90 degree bends and the monoblock it does go down significantly but it does look alright now. (i think)

 

[bigtallanddopey]

I feel kinda dumb decided to take apart the monoblock, there was some gunk on it, cleaned it and now I'm getting 1.95 GPM.
Running only rad its 2.75, so by adding 30 cm of tubing and 4x90 degree bends and the monoblock it does go down significantly but it does look alright now. (i think)

It's nearly always something simple. At least you found out the problem. Trouble shooting isn't easy on water loops especially when they use hardline tubing.

 

[GTXJackBauer]

I feel kinda dumb decided to take apart the monoblock, there was some gunk on it, cleaned it and now I'm getting 1.95 GPM.

I think it's safe to say that was your issue and hope it didn't spread to the rest of the loop.