ERV travails

Not open for further replies.


Well-known member
So some of you may remember some noises I made way back when about remanufacturing a Energy Recovery Ventilator in this house. It's actually a hybrid, featuring a ERV and HRV core, manufactured by Lifebreath as the TRV series way back in the 2000's. Anyhow, this one is configured as a fart fan, pulling air out of the bathrooms and putting fresh air back into the return ductwork on the AC system. Problem is, the control system in the unit was so user friendly that I am replacing it.

At the heart of the problem is that the motion-activated sensors in the bathrooms (24VDC on/off) only enable one response, full-on (!!!). That may be appropriate in some households with constant gastro-intestional distress issues, not so much here though.

Between toddling kids, cats, and so on, my wife got woken so many times that I had to disable this system. So, the first phase was to strip out the OEM PSC motor, control board and assorted junk. Then I added two hyper fans (ECM), one pulling air out of, the other pushing air into the house. These fans are speed-controlled using two optically-isolated resistance ladders with 8 inputs, i.e. 256 combinations (2.2k, 1.1k, and so on).

In order to balance the in vs. out air flow, the Lifebreath I&O manual wants you to use a manometer and adjust a damper to get to zero delta-P. Naturally, my original HVAC contractor didn't do that (One OEM once joked to me that most contractors use manuals solely as knee pads). I hope to achieve the same effect minus the manometer by using two Wind meters from Modern Devices to balance the air flow. So one blower gets up to speed, then the second one is ramped up until it achieves the same wind speed in the opposite direction.

Inside, a set of four DHT22's monitor the humidity and temperature of the incoming and exiting air streams. Eventually, I want to use this unit not just as a fart fan / TRV but also as a economizer. But that's some time away, just as hooking up a Nextion display / controller and a ESP12 for communications with the whole house DAQ. Today, the control daughter-board arrived from Laen, so hopefully a 3.2 will be running this ERV soon. Pictures to follow. Cheers!

PS: Seem to be much more successful with TRV so far than getting home fiber optic network running. Bummed I am, it appears that the structured cabling I had installed by an electrician way back when was sufficiently manhandled to break the fibers despite multiple layers protecting them (i.e. outer jacket cable bundle in addition to the jacket around the OM1 fiber pair). Out of the four fibers going into the office, only one reports a 23db loss (!!!) the rest of them have more than 50db loss (i.e. are likely write-offs). FWIW, I'm not blaming the electrician, any number of reasons could have led to that series of failures. I'll re-confection the office connectors, just in case, but I suspect fiber breaks elsewhere.
Last edited:
Re your blowers, rather than measuring the airflow, would it be better to connect the two ducts via a sample line, mount a flow or pressure sensor and adjust for zero cross flow/pressure? Course getting a read on static pressure like this can involve some interesting tweaking of your sample point to avoid reading dynamic pressure.
Reading OP it occurred that as you push air in the outgoing fan will move more air out as the vacuum dissipates - so could end up 'chasing your tail' and bringing in more than needed pushing directly out the running fan.

Reading the GremlinWrangler post - perhaps that is what he meant by reading dynamic pressure - though perhaps not.

Also if you over pressure the house with incoming air - any external duct could allow air to escape (kitchen fan, clothes dryer, chimney, furnace, other bathrooms) and that overpressure may not show on the subject fan once it hits steady speed. Of course not pushing in enough air could leave the house to make it up itself from unwanted sources. A single manometer could be the best solution - though alternate inlets and outlets could fool that too with under/over response.

PS: Constantin - not that I'll be doing fiber install - but how did you test it now? End use equipment or a tester? is there a good way to test the RAW material before and after install
Last edited:
Re your blowers, rather than measuring the airflow, would it be better to connect the two ducts via a sample line, mount a flow or pressure sensor and adjust for zero cross flow/pressure? Course getting a read on static pressure like this can involve some interesting tweaking of your sample point to avoid reading dynamic pressure.
Agreed that this is likely a better approach, but have yet to meet a delta-P sensor that I liked and that could be mounted a fair distance from the MCU (i.e. I2C need not apply). The wind approach is definitely crude by comparison. Averaging a couple hundred readings would likely remove all wobble / noise from the signal.
OK, here is a illustration of the OEM approach
Screen Shot 2017-03-12 at 12.47.12 PM.jpg
Note how they use one double-ended motor to turn two blower wheels. The two air streams do not physically touch - one HX transfers sensible heat via the HRV core (Al plates) and another transfers sensible heat via the ERV core (plastic). The two replacement blowers are in the same location as the motor, they are just two 8" Hyper fans instead of the PSC motor. So I doubt there will be any chasing /vacuum issues.

As for the fiber install, I used the Unicam Pretium stuff in the basement to finish the install. When equipment didn't sync up, I used a 650nm light source (LED or laser, not sure) to see if I could find breaks. It's pretty impressive how connectors that don't have a great termination will literally light up due to the light source. Ditto for the cabling, the light will shine right through the jacketing on the fiber itself.

Anyhow, only one of the fibers terminating in the office still had a "strong" signal using the visible light source. When I measured that fiber with my 850nm laser source + optical measurement device (they're sold as a kit), the measured loss came out to -23db, which is close to 10x what would have expected (after connector losses, <100' of fiber, etc.).

The other 3 fibers terminating in the office (altogether 2 sets of pulls) were extremely dim per the visual approach and worse than 50db loss per the measurement device. Thanks to closed cell foam insulation in the walls, unless the fix involves pulling the last 1-2' of fiber out of the wall and re-terminating that portion, I am cooked and will have to wait for widespread n-base-T adoption. That might allow me to get to 5 gigabits/s with the Cat5e in the wall today.
Last edited:
Nice picture - I was wondering where the energy recovery portion was if the fans were unique and distant points ... indeed that localization would minimize chasing like from a whole house fan. My house is a sealed box with no centralized air - just point sources of heat.

Interesting notes on the fiber issues and test hardware.
OK, back to the conversion. I started by removing the OEM motor, blower wheels and control system. This required the removal of the unit from above it's perch in the laundry room - some of the screws were mounted from the back of the unit. I modified the sheet metal to better reflect the new mixed flow blowers. Rather than feature small rectangular holes to match a volute that surrounds each blower wheel (per OEM), the holes are now "full size".
You can see the hyper fan lurking behind the hole. On the inside, you can see how I used a sheet metal collar to interface with the blower. What's nice about this collar is how it has double-sided foam tape to glue itself to the sheet metal. Once I mounted the fans, I used Al tape to seal them into the collars.
Here is one of four DHT22's. They will be used eventually to decide whether it's worth it to run the TRV as an economizer.
And here is one of two wind sensors. It uses a thermistor and claims a linear output.

I've been looking into differential pressure sensors in the 5" range and continue to be amazed by the pricing for this stuff. Little, if anything under $30, sometimes twice as much.
Last edited:
Part# source of wind sensor?

Click on the link above. It's a revision p wind sensor from modern devices. It still has a few improvement opportunities, i.e. a published set of ADC readings vs. measured wind speeds, information on temperature compensation, and potentially a AGND return path for the analog readings.
Good news is that I set aside a spare line (D30) which also happens to be a analog pin. So I will install a differential pressure sensor with an analog output from Sensiron to measure the delta-P. Thankfully, Sensiron offers a mounting base and even makes recommendations re: what connector to use. The irony is that the connector housing is inexpensive, but the four pigtals cost 10x!!!. I'm not bitter... really. Then, to connect it all, I am buying tubing from amazon plus barb fittings. All in, we're looking at $50 worth of parts. No wonder the OEM wants the installer to use a manometer!

What's nice about this sensor is that it operates in the right pressure range (i.e. +/- 0.5" WG) and that it's unipolar output is configurable two different ways based on whether a input pin is pulled high or low (square root vs. linear). Have a look at the datasheet, it's pretty nifty. Looks to me like this is a unit similar to their I2C offering, but they replace direct access to the I2C ADC interface with a single-output DAC. With the teensy ADC at 12 bits, I should have a pretty good idea when the in vs. out is balanced. It will be interesting to compare wind speeds and programmed speeds along the way, as I doubt that everything will be in balance all the time.
Last edited:
Anyhow, here is the current state of affairs. The TRV is back in its alcove, now I have to reconnect all the ducts, which is quite a challenge thanks to the small access hole I have to play with. I bought an inexpensive wind meter to compare the readings from the blue end of the unit (left) to the readings from the wind sensors. Plus, I have to run some more wires to hook up the new sensors. I'm using Cat5e for this task because its cheap, the distances are short, and the stuff is twisted.
Happy Ending after all.

Suspecting that the confectioning of the fiber optics by the electrician was the problem, I replaced the SC connectors in the office. All went wrong with fiber #1, it kept breaking until the nub in the wall was all that's left. But fibers 2-4 work wonderfully with negligible losses. Tested them (bright!) and then tested them again (less than 3db loss @850nm!), finally hooked the computer gear together... and it worked! Bottom line: I guess I get to redo the whole house re: SC connectors on the fiber optics. But if that's all it is...

Still waiting for the delivery from digikey for the differential pressure sensor.
I bought mine at Amazon, see the link above. About $200. I figure it's a good investment for a project like this one where every connection is likely suspect. However, the visual tester ($20) works very well also - if the visual signal is strong, the DB losses are low.

I guess it's good to have both, ie using the visual tester for a rough idea re losses and then using the calibrated source as confirmation.

Want me to sell the installation gear to you in a week or two? I also have the pretium install kit, which cost me $900 on eBay. New retail is $1700. It's the previous generation but it works well and I've confectioned over 25 connectors with it. The only problem one was in the office, of course!
Last edited:
Some of these optical issues are easy to diagnose. The light from a visual light source should be very bright inside a house when it arrives at its destination. So bright, that the outlet will glow from the inside.

From what I could tell, the electrician attempted a good install re: the connectors. They seemed to be stripped properly, i.e. all stubs were a consistent length, the cyanoacrylate layer was peeled (this is OM1 fiber). However, none of that will save you when the terminal is not installed properly. Here is one example after I pulled it out of the wall. Note the angle on the fibers.
Installing fiber inside walls without strain relief is likely going to make them fail consistently. Here are the same fibers after I pulled off the connectors. Note the "memory"! Light doesn't like sharp bends...
I cut the fiber jacket back quite a bit and then used the adjacent ethernet cabling as a strain relief (w/electricians tape). This will hopefully allow the fiber to maintain its minimum bend radius. Preliminary tests have been positive, i.e. every terminal so far has clocked in at less than 3db loss, which is pretty good considering that there are 2 patch cords and the actual fiber run to account for. I plan to verify the rest of the outlets before reselling the kit. It's another reason to not just get a good termination kit - invest in the tools to verify a good install!

Now back to the ERV.
Last edited:
The ERV project is moving forward. Tonight, I finished up the first test board and verified it's basic electrical operation. As you may recall, the Leviton motion sensors live off a 24VDC power supply. The Windsensor, revision P, from Modern devices needs ~9VDC or so for good thermistor operation, the Nextion display likes 5VDC, etc. So the first step is to take the 24VDC and turn it onto 10VDC for the wind sensor using a SWADJ power supply from Dimension Engineering. That output also goes into the DE-SW050 from Dimension Engineering, producing a nice stable 5VDC. 5VDC then goes into the Teensy 3.2 as well as the Nextion, etc. terminals.
IMG_8879 2.jpg
Soldering everything together went quite swimmingly. Attached the SWADJ first to a external power supply and adjusted the output until it hit 10VDC. Then inserted the DE unit and the Teensy started its happy blinkie dance. The VUSB-VIN link has been cut, a PMEG 4005 allows the Teensy to operate from either the 5VDC power supply or USB. The only head slapper was discovering that I managed to swap the VIN and VOUT on both switchmode modules (ARGH!). They still sorta fit. Feel like a dope though.
As you can see, I have room for a DHT22 in every quadrant of the ERV, along with wind sensors for both intake and exhaust to / from outside. The Delta-P differential sensor is likely best used when measuring from outside to indoor ambient? For the two sensors at the far end of the ERV I may end up using two 3.3VDC power supplies running off the 10VDC bus, just to ensure that the power supply out there is stable. 3.3V appears to be on the ragged edge of what the DHT22 will endure.

The opto-isolators are for the two fans, the chips are removable. The terminals for the bathroom zones and the wind sensors are obscured by the switchmode power supplies. Teensy USB port accessible w/o issues. Both Teensy and board are OSH purple. Happiness.
Last edited:
Not open for further replies.