そんなことになっているんですね... まだ不安定なのかもしれないですね。
こちらでも見られるようになりました。x.com/home も見えるみたいです

そちらでは戻ったんですね。よかったです。
こちらはまだ戻ってないみたいなので地域差があるかもしれないですね...

みたいですね...

Intel Macも対応はしているように見えたのですが、メモリ量の要求が厳しそうです

Blue is a comparator output.
The second figure is the waveform when the current direction is the same as one of the previous current application.
The third one is when the current direction is opposite to one of the previous.
As we can see, the period of the low level output is different.

Here are the schematic of the sense part and observed waveforms. There are no strong reason to use 1k ohms in the schematic, and there may be more better choice. Two 100 ohm VRs are for tuning of sensitivity.
Orange is observed waveforms between the two ends of the red wire in the picture.

I used LM393 as the comparator. Arduino software monitors the comparator output. It is critical part in timing, and I wrote it with inline asm. Temperature compensation seems to be necessary.

Here is the schematic of the driver. I used ULN2803 instead of full bridges. Though the maximum rateting for a channel is 500mA, it exeeds the value a little. But, the datasheet of ST's chip tells that it permits 600mA, I think.
The sense amp is based on corememoryshield.com/report.html

Because there is no instrument to check BH curve, candidates were picked up with parameters in data sheet.
Parameters of material 25 can be found in following datasheet. But, BH curve for the material is missing...

www.laird.com/sites/defaul...

I tried other cores. I verified Fair-Rite 5952020201 cores working in 2x2 configuration. But I have not build 4x4 configuration with the Fair-Rite cores. PCBs for 4x4 is under production.

「基本」→「言語」→「コンテンツの言語」のEnglishを外した状態で「ホーム」の画面を確認してみたのですが、フォローしているアカウントの投稿は日本語も英語も両方とも見えているみたいでした。