Page 34 of 108

PIP-4048MS inverter

Posted: Thu, 31 Dec 2015, 04:05
by coulomb
Fan "Rectification"

This post is for those who agree with me that the fans in any electronic equipment, if they act vertically, should blow air up from below, not suck out from below. Performing this modification will likely void your warranty, despite the fact that in my opinion, it will improve the cooling of the unit.

[ Edit: The other consideration is that dust tends to collect where the fans draw air into the unit. This is more readily accessible at the bottom of the unit, where the cable access cover can be removed with just two screws. ]

The top cover comes off easily, but you'll have to sacrifice part of the sticker with the year of manufacture. This is what you need to aim for, in order to be able to access the bottom two screws of the left hand fan; they're right behind the serial PCB:


NOTE: It's been shown that the following is the hard way to access the fans. I leave the following for historical interest and for the reader's enjoyment at my expense. Take out the serial board for much easier access.

If you don't get the plate holding the two fans this high up, then you'll never be able to remove and replace the bottom two screws on the left fan. Well, unless you have an amazing Z screwdriver that has a very short straight part. The two screws holding the fan plate to the box sides are easy to remove, but unlike earlier models, the PV input terminal block is now holding the plate firmly in place. The terminal block is screwed to a seat, as shown below:

Image     Image     Image

The second photo shows the pegs that together with the screw and the tight fit hold the PV terminals in place. The third photo shows the three screws that you need to remove: one in the middle that screws into the seat, and the two holding the straps that connect the PV inputs to the SCC PCB. The other end of the straps is soldered to the SCC PCB; it would be a lot of effort to undo those, and these is no need. The terminal block comes out with a bit of wiggling and choice words.

At this point, the fan plate still doesn't come up, because of the two toroids marked 1 and 2 on the photo below, trapped under the printed circuit board marked 3:


By pushing the fan plate towards you, you can carefully work the two toroids out from under the printed circuit board. Finally, the fan plate can come up. Using a full sized #2 [ edit: was #3 ] Philips head screwdriver, the fan screws can be removed.

Oh, I've neglected to mention another test of dexterity and patience. There is a small cable tie at the bottom of the left hand fan, which restricts movement of the right fan. Using long nosed pliers, I was able to pull the cable for the right hand fan through the cable tie, but not far enough to be able to rotate that fan. I found the best way to cut the cable tie was to use a small flat bladed screw driver to rotate it slightly so that the bulge on the cable tie was facing away from me; I was then able to cut it with medium sized side cutters from above.

It is then a matter of taking the fans up, rotating them about the vertical axis (so they now blow inwards rather than outwards), and putting them back. You will be cutting new threads into what was the back of the fans (now the front), so that's another reason you need good access to all eight fan screws, and the need for the large #3 Philips screw driver.

Manipulate the toroids back under the printed circuit board, carefully as before. Replace the PV terminal block, screw the fan plate back, and replace the lid.

Edit 1: in retrospect, it might have been easier to remove the serial port board. You could cut the new thread in the fan plastic while the fans are out of the unit. I may try it that way if I decide to replace the fans with quieter ones.

Edit 2: Weber just reversed his fans without having to remove the metal plate that they are mounted on at all. Removing the serial port board is definitely the way to go. He may have rotated one of the fans 90 degrees to make the cables fit. He didn't even have to cut the troublesome cable tie.

PIP-4048MS inverter

Posted: Fri, 01 Jan 2016, 02:55
by coulomb
I'm playing around with my inverter, watching the temperature with various loads, and I see that the temperature sits at 44-52°C now. On 52.02 it would make a lot of noise (with original fans) but keep the temperature around 32-35°C.

The other thing I notice is that the lower temperatures happen with *higher* load. So there must be a feed forward term there ("load is 40% - set the fans up higher, don't wait to measure the temperature rise, we know it's coming"). The lowest temperature I measured was 41°C immediately after a moderate load (2.3 kW) came off.

With ambient load (100-120 W), I'm getting 47-51°C; with the microwave running (1.8 kW total), I'm getting 44-46°C; with a 2.3 kW total load, 42-43°C. At this ambient temperature (24°C), it's slightly over compensating. The actual junction temperatures will be higher than measured, so that's good.

So far, the stock Adda fans have been quite quiet. Only once did it run the fans at a moderate huff, but only for less than a second, and nowhere near as loud as with the older firmware.

I think eventually I'd like a lower average temperature, but for now, the stock fans and fan algorithm will do.

[ Edit 9/Jan/2016: I had the PIP horizontal, which really invalidated the test. I noticed that the front cover was getting warm; with later tests and a vertical PIP, this did not happen. ]

PIP-4048MS inverter

Posted: Fri, 01 Jan 2016, 17:17
by offgridQLD
I assume you didn't have any PV connected to the SCC at the time when measuring inverter temps under load.

I would be interested to see what happens with say 100-200w base load on the Inverter at the same time that the SCC is charging the batterys at say 50A for a while.

Perhaps high temp would then trigger the fans.

My experience is that inverters in offgrid house applications tend to spend most of the time doing not much (idle loads) Typically large load times are all about the kitchen and laundry. You have the morning breakfast rush hr loads and the same for lunch and dinner. With the odd load of washing or ironing thrown in.

A good 12hrs or so over night they just idle along with the odd little spike from the fridge now and then. A large portion of the day time they are doing much the same.

Perhaps having the fans running gently under light load would be nice just to lower that idle/base load temp below 40C to help with component longevity.


PIP-4048MS inverter

Posted: Fri, 01 Jan 2016, 17:54
by coulomb
offgridQLD wrote: I assume you didn't have any PV connected to the SCC at the time when measuring inverter temps under load.
No, I forgot to mention that.

But I am today. It seems that even with very light PV charging (e.g. 2 A into the battery), the fans run harder and the temperature targets 42-44°C. [ Update: but that doesn't hold up now; even 7 A of charging sometimes sees the temperature peaking at 51°C. ]
I would be interested to see what happens with say 100-200w base load on the Inverter at the same time that the SCC is charging the batterys at say 50A for a while.
I only have two strings (4 panels) on the PIP at present, and this morning has been very cloudy. But I'd say it would continue to target the low 40s in degrees.
Perhaps having the fans running gently under light load would be nice just to lower that idle/base load temp below 40C to help with component longevity.

Yes, I think that would be good. But as the firmware is now, at least during the day, the temperature will be eight [ edit: was ten ] or so degrees cooler. The fans running slower at night will at least reduce the battery-only load by a very small amount.

Update: with around 15 A solar charging, it's maintaining a solid 42°C.

Update 2: there were many flaws with this test; see this post.

PIP-4048MS inverter

Posted: Fri, 01 Jan 2016, 18:34
by coulomb
When sending the same command to a PIP inverter-charger over and over, especially one with a many-keystroke CRC sequence like QPIGS, it's convenient to use a program like AccessPort, which allows you to send a sequence specified in a mixture of text and hex codes with a single mouse click. It can be downloaded here: (Windows only; doesn't mention Windows 8 or 10 but works for me on Windows 10.)


[ Edit: there is also an auto-send capability. I find that an 8000 ms (8 second) delay is good for the QPIGS general status command. ]

PIP-4048MS inverter

Posted: Sat, 02 Jan 2016, 19:18
by dockarl
Speaking as a person that's spent my fair share of time deep in the bowels of code for embedded systems like this one I'd have to say - I'm sure everyone realises - what Coulomb and Weber have achieved by reverse engineering the firmware here is 'NO SMALL FEAT' - not by any stretch. There would be very few people around with the skills to do that - definitely beyond my pay grade. I can't even begin to imagine the challenges they would have encountered.

Congrats to both of you - very impressive achievement - and thanks very much to both of you for putting my blown inverter to good use as a 'testbed' during the reverse engineering - with the cool side effect that it's now working again :)

Thanks guys - very much appreciated. HNY.


PIP-4048MS inverter

Posted: Sun, 03 Jan 2016, 15:50
by paulvk
Coulomb have you turned the fans around it would be good to see the difference if any ?
Also if you get a flexible driver you can remove fans without taking the units apart ... B0099QV18Q

[ Edited Coulomb: linkable URL ]

PIP-4048MS inverter

Posted: Sun, 03 Jan 2016, 20:30
by coulomb
I don't think that one would be nearly flexible enough for two of the fan screws. [ Edit: this was while I wasn't considering taking out the comms board. ] There is only about 15 mm clearance to the back of the serial comms board.

PIP-4048MS inverter

Posted: Sun, 03 Jan 2016, 20:35
by coulomb
paulvk wrote: Coulomb have you turned the fans around it would be good to see the difference if any ?

Yes. But I foolishly didn't do measurements before I turned them around. Weber has promised me to do the before and after measurements on his recently ordered PIP. It should arrive in a few days.

PIP-4048MS inverter

Posted: Mon, 04 Jan 2016, 14:47
by Jdp
Hi all. Mike has been helping us here in South Africa. These inverters are a bit new to us. I did not get any useful stats out of the software that came with it. So I decided to build my own, thanks to the protocol this forum provided. People over here in SA saw what I did and asked me to make it available for them to use as well. So I spent about 2 months of my live sitting till 2 in the morning to build this software. So if you are interested there is a 15 day free trail. It works the best with a Victron BMV-702 as we all know we can not trust the battery values that come from the inverters.

Link to the site you can get the free trail

Here is the link to my solar setup on-line.

All the best.


PIP-4048MS inverter

Posted: Wed, 06 Jan 2016, 17:50
by PlanB
Dave/Mike, is the typo fix you made after your day with the monolith incorporated in the download posted a few pages back? Would like to avail myself of the improved low voltage cutout for lithiums.

PIP-4048MS inverter

Posted: Wed, 06 Jan 2016, 18:23
by coulomb
Yes, all files with the typo bug are fixed.

PIP-4048MS inverter

Posted: Sat, 09 Jan 2016, 20:26
by coulomb
paulvk wrote: Coulomb have you turned the fans around it would be good to see the difference if any ?
Ok, now that Weber's PIP is here and it's Project Day, I can rectify my lack of a scientific comparison.

We set up his PIP almost vertically (another mistake I made); ambient temperature was 26°C. We loaded it with 3.8 kW before and after the fan reversal. PV panels were connected; it was a cloudy day with intermittent full sun.


To cut a long story short, with the fans in the original direction (blowing out of the bottom of the unit), at 3.8 kW load, it quickly reached 57°C. With the load reduced to 100 W (to compare better with my results), the temperature actually increased to 58°C, and remained there for at least 3 minutes. With zero load, it fell to just 57°C.

To prove this wasn't some artifact, we re-connected the 3.8 kW load, and the temperature went down to 56 then 57°C.

After the fans were reversed, with the 3.8 kW load, the highest temperature reached was 45°C. With the 100 W load, the temperature went up to 46°C (with the fans off), then reduced to 41°C after 5 minutes.

We were pretty slack about checking the PV charging power. With the original fan direction, the two measurements we noted were 300 W and 1200 W. At the end, with the 41°C figure, the solar input was about 300 W. So we waited for some sun; it rose to 1.2 kW. This only ran the fans a little harder, and reduced the temperature to 40°C.

It seems that with the fans blowing down as standard and 100 W of load, the main effect of the fan was to precisely cancel the benefits of natural convection Image .

In summary, the fan direction change gives a 12°C improvement at high loads. At 50% load, there is an improvement, but only four degrees.

PIP-4048MS inverter

Posted: Sat, 09 Jan 2016, 20:35
by solamahn
Good. I will continue to turn all my fans upside down pre installation. I have some msd and mst coming next week.

PIP-4048MS inverter

Posted: Sat, 09 Jan 2016, 21:26
by paulvk
It is good to have real data.
Today with temps here at 27deg I had the air con running and then boiled the kettle this had the unit at 3200watts for about 4 mins .
As the top of the inverter heat sink had gone past 42deg the two 125mm fans mounted on the sides at the top were running (my add on external), with the load up the two internal fans were running almost flat out the temperature stayed at 40deg.
This is the older unit with heat sink at the top there was also 48amps from the solar panels.
Now I have my temperature probe from the controller (similar to this ... pCYnLGZCpo) at the top of the inverter heat sink.
How are you measuring the temperature, the inverters sensor on my unit appears to be at the bottom so blowing air at it from the bottom would maybe not give good results.

PIP-4048MS inverter

Posted: Sat, 09 Jan 2016, 23:52
by coulomb
paulvk wrote: How are you measuring the temperature
This is using the QPIGS command, with AccessPort. That's presumably the same measurement that the firmware is using.
The inverters sensor on my unit appears to be at the bottom so blowing air at it from the bottom would maybe not give good results.
An interesting question. It seems to take the maximum of about 3 measurements, and I think one of those is bolted to the heatsink.

Weber seems to recall a bead thermistor physically under the main transformer. It may have been lower than the main transformer, which means without the reversal it would have gotten the heat from that transformer, but with the reversal, it might not be registering that heat.

It would be good to know where the sensors actually are. I can check on a spare board when I get home, for the 2013 model.

PIP-4048MS inverter

Posted: Sun, 10 Jan 2016, 05:13
by coulomb
coulomb wrote: Weber seems to recall a bead thermistor physically under the main transformer.
He remembers well:

Image     Image

The above photo is of a 2013 model, kindly donated for spare parts by forum contributor dockarl. The transformer doesn't look much like a transformer from this angle; most of what can be seen is something shiny on the side of its ferrite core. It may be plastic tape, varnish, or a lacquer coating.

NTC.Txt probably stands for Negative Temperature Coefficient thermistor, Transformer Temperature. You can see that there is no thermal connection between the transformer and the tiny thermistor. Indeed, the transformer appears to be thermally isolated from the printed circuit board by the white plastic material. (Some heat would conduct to the PCB tracks by conduction along the leads.) So it must sense by radiated heat, or conduction through air. I imagine that radiated heat would dominate.

The thermistor is towards the bottom of the unit when mounted normally, so that indeed blowing colder air from below may in fact change how the transformer's temperature is sensed. This is not something I'm an expert at, so I don't know whether this is serious or not.

There are also two more standard thermal sensors of some sort bolted to the two main heatsinks. So there are at least four temperature sensors: two on the heat sinks, one or more on the Solar Charge Controller, and this thermistor.


[ Edit 1: I decided the shiny surface could be plastic tape as well as varnish/lacquer. ]
[ Edit 2: I found that the transformer sensor read 5-6 °C cooler 4 seconds after the fans came on at 30% after a 60-160 W load. It "bounced back" 4 seconds after the fans came back on. This suggests to me that reversing the fan air flow direction may be causing the transformer to heat up some 5 °C more than it would have with the standard fan direction, under certain unusual conditions. ]
[ Edit 3: Added heatsink sensor photo. ]

PIP-4048MS inverter

Posted: Sun, 10 Jan 2016, 16:21
by coulomb
In answering a post on the South African Power Forum, I came across the Q1 command. I so wish I was aware of this command yesterday when Weber and I did the fan temperature experiments. (So you're a day late, guys! Image )

It seems to be tailor made for investigating the effectiveness of the fan control system! Like most of the well over 100 commands that the PIP responds to, this one is undocumented. (Even some of the documented commands have extra fields; for example 4 at the end of the QPIGS command. For example, you can see the PV power was around 990 W in my Access Port post above; it's the second last field. The string of eight ones and zeroes ("Device Status") is the last documented field (at least, with the documentation I have, with 20140822 in the file name).

The Q1 command (CRC 1B FC) seems to send back the following:
(        All command responses start with the open parenthesis character.
AAAAA    Local inverter status (first field). This seems to be a bit field,
         commonly taking the values 0x3809 or similar (shows as 14345 decimal).
         Edit: this seems to become a count in seconds till the end of CV
         (absorb) charging, in firmware version 72.70.
BBBBB    ParaExistInfo first field. This always seems to be 00001, even with
         no parallel card installed.
         Edit: this seems to become a count in seconds till the end of float
         charging (when it will start CC (bulk) charging), in firmware version
CC       SccOkFlag. I assume that 1 means the SCC is powered and is
DD       AllowSccOnFlag.
EE       ChargeAverageCurrent. I'm not clear on what chargers are included.
FFF      SCC PWM temperature, in °C. From global variable wSccPWMTemp.
GGG      Inverter temperature, in °C. Presumed to be from the AC heatsink.
HHH      "Battery temperature". It seems that this must be the temperature
         reported by a sensor on the battery to bus inverter heatsink.
III      Transformer temperature. It's the result of calling _wTempDegreeTxt().
         Presumably also in °C.
JJ       GPADAT bit 13 (i.e. whatever is connected to general purpose I/O line
         GPIO13). It is presumably related to temperature and fans.
KK       FanLockStatus. I'd say 01 means fans are locked, 00 means not locked.
LLL      FanPWMDuty. No longer used. Always 000.
MMMM     "FanPWM", but is actually speed in percent. 0000 represents off, and
         0100 represents 100% duty cycle (flat out). However, on start-up, this
         value goes to 0100 without the fans roaring. 0030 (30%) seems to be
         the lowest speed, quite quiet. At 42% load, the fans went to 42% 
NNNN     SCC charge power, watts. This is one of the changes to firmware
         version 72.40 that is not present in version 52.30. In 72.40, the
         result of the call to _swGetSccChgPower() is divided by 10; in 52.30
         is is displayed as is. I suspect 52.30 would have displayed tenths of
OOOO     ParaWarning. Presumably, some warning bitfield related to paralleled
PP.PP    SYNFreq. Wild guess: frequency of inverter after synchronising with
         the mains input.
QQ       Inverter charge status. This will likely be 10 for no charging, 11 for
         bulk stage, 12 for absorb, or 13 for float. However, bulk stage will
         usually report as 12, same as absorb. I don't know what the signif-
         icance of the leading "1" digit is; I've always found it to be one, but
         the firmware calculates this value modulo 10 (stripping off the tens
         digit) a lot of the time.
These 17 fields are followed by the usual checksum and carriage return.

So you can see that an experiment could possibly be designed to find out whether the transformer temperature measurement is being affected by the fan air flow direction. However, it would probably require either some guesswork, or access to the transformer while the unit is running. The latter is likely not practical on a 2015 model machine.

[ Edited 11/Jan/2016: changed several descriptions after actually examining the results (no load) ]
[ Edited 17/Dec/2016: Inserted PWM temperature field (FFF); moved other fields down (so now ends as QQ instead of PP ]
[ Edited 14/Jan/2017: Guess on the meaning of "Battery temperature" changed. SynFreq now a wild guess. ]
[ Edited 3/Feb/2017: Amended guess of first two fields for firmeare version 72.70 ]
[ Edit: Always found it on -> Always found it to be one ]

PIP-4048MS inverter

Posted: Sun, 10 Jan 2016, 20:29
by paulvk
Be nice to find a command that sends back a list of commands the unit responds to!

PIP-4048MS inverter

Posted: Mon, 11 Jan 2016, 16:48
Hi All,
I have recently purchased a 4048 inverter. Just wanted to ask opinions on if I should change the caps on the DC side now?.....Thanks in advance!

PIP-4048MS inverter

Posted: Mon, 11 Jan 2016, 23:28
by paulvk
Is it under warranty?
If it is why void the warranty.
I have units here working 24/7 for over 12 months have had no problems with the caps and as they are still under warranty (2 years on my ones) I will wait till that ends.
I have lead acid batteries connected to the units.

PIP-4048MS inverter

Posted: Mon, 11 Jan 2016, 23:41
Yes, the unit is new, so still under warranty. Waiting until the warranty ended was my thinking also, however, I don't want it breaking down if it's a simple case of changing them now to help solve an issue I would be happy to do that. I was a little concerned by the quality of the caps, reading about other peoples experiences.

PIP-4048MS inverter

Posted: Mon, 11 Jan 2016, 23:58
by weber
It is possible that in recent models the caps have been upgraded, and/or that voltage overshoots are better controlled. I'm unwilling to disassemble my new unit enough to find out.

PIP-4048MS inverter

Posted: Thu, 14 Jan 2016, 15:01
by offgridQLD
Am I correct in thinking if I connect the PIP to a battery with PV connected and switch the inverter of with the rocker switch. The charge controller will still wake up each day and charge the battery even if the inverter is switched off?

I just want to recharge a battery but don't want the 48w inverter base load on the battery 24/7


PIP-4048MS inverter

Posted: Thu, 14 Jan 2016, 16:17
by paulvk
Yes it will charge the battery when the sun comes up.
I have a unit sitting here that only has batteries connected and plugged into mains on standby (inverter off) as I want to write a program on a micro to talk to it and measuring the battery drain with no mains I see 250ma so you will have a drain on your battery I expect the same until the sun comes up!

Note Also that I measure the power factor to be .11 at the input to the inverter!