Issues with transformer inductance

[LucaStefanRusu24]

Hello,

While testing the transformer, I am finding that the inductance values exceed the 5% error margin that the Build Guide mentions. I first tried manufacturing 2 transformers initially when the build guide was not updated and only contained 28 turns. These were tested using an LCR meter and the following values were obtained:

The values, especially for the primary, deviated by quite a large amount, but I assumed this was mostly due to winding errors. I then wound two other transformers with 29 turns and while following the winding video provided. I tested these new transformers again but still got similar values:

This indicates one of two things: 1) the winding was not the problem or 2) the windings were done just as badly on both sets of coils. Assuming the first option is true, I am trying to debug what else could make the transformer have a such great deviations in its inductance values, so I would like to know if there are any suggestions on what could possibly be the issue or if it is indeed faulty winding (aka skill issue) on my side.

A few differences to note about our transformer:

• The N97 core was shipped in a much larger size than the N27 cores, and could not be used as the 2nd half of the transformer core, so we used two N27 cores for the transformer instead. This I would imagine would cause some deviations in the values but not as large as the ones displayed

• The wires we used for the winding were Litz wires, 120x0.1 for the primary and 600x0.071 for the secondary, but they were not the ones from the eBay link in the build guide, as those were not rated to 600V, so we instead got wire samples from Elektrisola which said that the wires should specify our requirements, but did not give an exact breakdown voltage.

Any suggestions are welcome!

[Rootthecause]

Hi Luca,
thanks for providing the measurements and good breakdown of your thoughs.
The Litz wire should be fine - good thing that you got the (extra) rated litz from Elektrisola :)

Your issue is probably the core. The N27N core is the one with 1 mm airgap. If you use two of them, it's 2 mm airgap (should be only 1mm), and thus the magnetic resistance in the core is higher, resulting in a lower permeability seen from the coil and thus resulting in a lower inductance value. So using an ungapped core for the other side should resolve this. In case you somehow have a ungapped N27, N87 or N97 any of these should work as well, but I would recommend N97 for slightly better efficiency. Let me know, if that resolved it :)
Also make sure that your cores are properly pressed together and no other gaps are present, except in the middle.

Just wondering if anything went wrong on my side, but I checked that I had the correct core in my files. What core did you get instead? E43 ? (Note: E39 means the width of one core is 39 mm).

Oh, and one small thing: Inductance is measured in Henry, not Farad :)

[LucaStefanRusu24]

Thanks a lot for the response! I dont think anything went wrong on your side, it was probably an error on my side or my department manager’s side when ordering the cores and before ordering more cores we tried to test with the ones we did get to see the results since our cost manager gets a bit stingy about buying stuff and it takes a while :). I imagined that the 1mm difference in air gap could have an impact on the inductance in the end but I didn’t know how big that impact was. I will pass the information to my department manager and ask to get the N97 coils. And yes indeed, inductance is measured in Henry, I tend to get those mixed up a lot for some reason. Thank you for the help! Side note, when starting out the project I was stuyding the conversation you had with the scrutineer from FS Czech on the repo and saw one of the issues they raised was the fact that the transformer wires being soldered directly to the board was not rule-compliant. I am now trying to implement a design that uses press-fit connectors to connect the transformer ends to the pcb. Will let you know if that actually works, might be something to think about as a fix. If you think of any concerns about this idea feel free to let me know as any input is helpful.

…

On Tue, Jun 16, 2026 at 23:21 Rootthecause ***@***.***> wrote: Hi Luca, thanks for providing the measurements and good breakdown of your thoughs. The Litz wire should be fine - good thing that you got the (extra) rated litz from Elektrisola :) Your issue is probably the core. The N27N core is the one with 1 mm airgap. If you use two of them, it's 2 mm airgap (should be only 1mm), and thus the magnetic resistance in the core is higher, resulting in a lower permeability seen from the coil and thus resulting in a lower inductance value. So using an ungapped core for the other side should resolve this. In case you somehow have a ungapped N27, N87 or N97 any of these should work as well, but I would recommend N97 for slightly better efficiency. Let me know, if that resolved it :) Also make sure that your cores are properly pressed together and no other gaps are present, except in the middle. Just wondering if anything went wrong on my side, but I checked that I had the correct core in my files. What core did you get instead? E43 ? (Note: E39 means the width of one core is 39 mm). Oh, and one small thing: Inductance is measured in Henry, not Farad :) — Reply to this email directly, view it on GitHub <#17?email_source=notifications&email_token=ALMCSSNBUYOCBOH22P7TDNL5AG25NA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGI4TINZRUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVRTG633UMVZF6Y3MNFRWW#discussioncomment-17329471>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ALMCSSMBRSLHV6E7QNBHGG35AG25NAVCNFSNUABIKJSXA33TNF2G64TZHM2TIMRWGY4TOOBYHNCGS43DOVZXG2LPNY5TCMBSGY4DMNBWUF3AE> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/ALMCSSPMWVVZ5NH53UUKRBT5AG25NA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGI4TINZRUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVJTG633UMVZF62LPOM> and Android <https://github.com/notifications/mobile/android/ALMCSSIMYOID7GU6G6QEPAL5AG25NA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGI4TINZRUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVZTG633UMVZF6YLOMRZG62LE>. Download it today! You are receiving this because you authored the thread.Message ID: ***@***.***>

[Rootthecause]

Ah okay :) 

I thought I did write it down here as well somewhere, but it seems I didn't. You can find the answer to the rules question regards solderin the hf-litz to the PCB in the scrutineering support presentation on slide 14:

I might respond to that issue in the future, but I still havn't gotten any other answer from the Czech Scruti so far.

The problem with HF litz wire is, that the enamell on every litz must be removed. The only proper way to do this is by tinning. So even if you do press-fit, it must be tinned before. We got the info, that tinning is the appropriate way of installing the HF litz from Elektrisola. If you would press-fit the tinned litz, it might become loose when the solder gets soft. So it doesn't solve much.