Henning Paul DC4HP

Electronics tinkerer, DC4HP, Dr.-Ing. ET/IT, Ex-PostDoc, multi-instrument dilettante, cornucopia of superfluous knowledge, Xennial, husband, father. He/him.

Public Key

npub1akzwlted6r8t0ap62ax7fssxzl20tvxwnw9f9hls02zeccfsjyfq98cn50

NIP-05 Address

hennichodernich_at_radiosocial.de@momostr.pink

Profile Code

nprofile1qqswmp804ukapn4h7sa9wn0ycgrp0484kr8fhz5jmlc84pvuvycfzyspz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dsdskq8w

Publishing to

relay.nostr.band

relay.damus.io

nos.lol

Author Public Key

npub1akzwlted6r8t0ap62ax7fssxzl20tvxwnw9f9hls02zeccfsjyfq98cn50

Show more details

Last Notes

2026-04-15 05:12:57 UTC

- reply

really "brand new" or showroom model?

2025-11-13 20:11:09 UTC

- reply

Yes, just make sure to follow the layout guidelines that are given in the datasheet of the switcher.

2025-11-13 20:08:22 UTC

- reply

3.3V to 1.2V is okay (0.63 watts), going straight from 5 to 1.2 would have been problematic.

2025-11-13 19:49:31 UTC

- reply

I have just skimmed the datasheet, but you should use MLCCs as input and output capacitors (see chapters 8.2.2.2 and 8.2.2.3).

2025-11-13 19:20:05 UTC

- reply

To be honest, the only valid application to use a high-current linear regulator in 2025 is when you need a low-ripple voltage for your analog circuitry. In that case you use a switching regulator to step down to Vout-Vdrop and put the LDO regulator behind that. By doing so you can reduce the voltage ripple of the step-down by the LDO's PSRR.
Disclaimer: I design (and more often review colleagues') analog electronics for a living.

2025-11-13 19:16:58 UTC

- reply

Just don't pull that much current when the voltage delta is that large...
Low drop regulators need 500mV, maybe 700mV drop to operate. So you add series power resistors to the input that at maximum current Vout+Vdrop is on the input pin. This doesn't improve efficiency, but spreads power dissipation among components.

2025-11-13 19:04:04 UTC

- reply

If you look at the graph, (5-3.3)V*1A=1.7W is still out of spec, and (5-1.2)V*0.3A=1.14W is only ok if the cooling pad is large enough. (Note that the graph assumes 2oz copper.)
Unfortunately for the 1117s the tab isn't GND, so you can't use the GND copper pour as heatsink.

2025-11-13 18:41:25 UTC

- reply

Exactly.

2025-11-13 18:36:20 UTC

- reply

5.7 watts on the 3.3V regulator and 4*1.3=5.2 watts on the 5V regulator... Exceeds the maximum allowed power dissipation by far.
https://radiosocial.de/system/media_attachments/files/115/543/830/554/781/605/original/6daf2a80ac9402f4.png

2025-11-13 18:26:26 UTC

- reply

Do you plan to use a 9V power supply? Toasty!

2025-11-04 18:58:24 UTC

- reply

Ah, like "Hengst"!

2025-11-04 18:53:35 UTC

- reply

I wonder how Danish "hest" mixes into this.

2025-06-27 16:59:35 UTC

- reply

Phew, if you know how many people fled from the GDR before the wall was built (and even afterwards, many paying with their lives)...

2025-06-27 15:31:26 UTC

- reply

I don't know if this disqualifies her for the job. She never met him, not even her father ever met him.
You think maybe there's something evil in her genes?

2025-01-02 16:37:13 UTC

- reply

I just made some measurements with an EPCOS/TDK B62152A1X1 binocular core. It has an A_L of 330nH+-30% (assuming full turns around the center leg). 4 turns measure 5.74µH, so that figures. A single wire through one hole measures 236nH.
1/n
https://radiosocial.de/system/media_attachments/files/113/759/722/787/499/810/original/2393056947d63da0.jpg
https://radiosocial.de/system/media_attachments/files/113/759/723/146/246/529/original/7c01ec1754b81780.jpg

2024-12-31 13:19:01 UTC

- reply

This is obvious when you think harder about it, using that rule of thumb from the ham radio course, an open loop, i.e., Z->\infty would have been transformed to Z->\infty on the primary side as well, but instead we see the no-load inductance, because the assumption Z << j\omegaL1 no longer holds.
That also implies that a transformer becomes "better" at transforming impedances the higher its no-load inductance is.

2024-12-31 13:16:02 UTC

- reply

This only becomes tractable if we assume an ideal transformer, i.e., M12^2=L1*L2. Toroids are not so bad in this respect, but still an idealizing assumption.
Deriving this helped me cut the Gordian knot I had in my brain since the ham radio course. A transformer will only transform impedances within certain bounds, i.e., the load impedance has to be significantly smaller than the no-load impedance. Only then, Z is actually multiplied with L1/L2, i.e., n1^2/n2^2.
https://radiosocial.de/system/media_attachments/files/113/747/606/207/763/446/original/677183225ccb632e.png

2024-12-31 13:10:37 UTC

- reply

Now the impedance-transforming thing, the cargo cult of all radio amateurs.
In general, you see the folloing impedance on side 1 if you terminate side 2 with load Z:
Not so nice, obviously.
https://radiosocial.de/system/media_attachments/files/113/747/597/901/092/264/original/f54a3f868f35e161.png

2024-12-31 13:07:54 UTC

- reply

The stray inductance is what you have to put in series with an ideal transformer (of reduced inductance) to model a nonideal one (the overall no-load inductance has to be the same).

2024-12-31 13:05:04 UTC

- reply

And because I don't want to promise things I can't keep heres how you get these properties: The no-load inductance is trivial: Without load, I2 is 0, therefore the current controlled voltage source on the primary side is 0, too. You only see L1, this is the no-load inductance.
Now the short-circuit case:
Depending on the coupling k, you either fully see L1 (k=0, no coupling), or you see a short circuit (k=1, ideal transformer).
https://radiosocial.de/system/media_attachments/files/113/747/572/233/579/279/original/91e853d5b70ddabb.png

2024-12-31 11:36:45 UTC

- reply

This is intended as a general thought piece, not a reply to your post, but I don't know where else to attach it in this thread, so here we go:
This is the general equivalent circuit of a 2-winding transformer. Nonideal, but linear. Everything we assume about them can be derived from it: no-load inductance, short-circuit inductance, stray inductance, impedance transformation properties.
https://radiosocial.de/system/media_attachments/files/113/747/213/216/667/367/original/e7714b4753849243.png

2024-12-30 22:41:15 UTC

- reply

The blue squiggles represent the stray inductance, larger if coupling is worse.
I don't see how the lower left branch should be somehow in series with the output on the right hand side.
https://radiosocial.de/system/media_attachments/files/113/744/176/129/786/701/original/77b8d6a14466de3a.png

2024-12-30 22:31:37 UTC

- reply

I still don't get what you mean by "choke". A series inductor? Between where and where? Or a common mode choke?
A transformer is all about currents. A current in any of the 3 windings will induce a voltage in every of the 3 windings (proportional by M_ij or L_i respectively}. This doesn't do anything per se, as long as this voltage doesn't cause a current (canceling the voltages in the other windings per Lenz's law).

2024-12-30 21:04:30 UTC

- reply

The inductance towards the rail (because the MOSFET is high impedance)?

2024-12-30 18:52:10 UTC

- reply

You basically have 3 windings in total, but the coupling between the tubes is much lower than between each tube and coil, if I'm not mistaken.
Therefore, not much voltage is induced in the branch of the inactive MOSFET and the effect of its body diode can be ignored.

2024-12-27 22:17:32 UTC

- reply

Having that much disposable income for hobby purposes deffo is an outlier.
I know that others buy yachts and classic cars, but still.

2024-12-21 18:25:23 UTC

- reply

Why the exotic low-noise BC549 and not the general-purpose BC547? It's a digital circuit after all.

2024-12-20 14:07:14 UTC

- reply

OC outputs are slooooow compared to push-pull outputs in the presence of parasitic capacitances.

2024-12-18 06:11:06 UTC

- reply

I'm currently struggling with dumping an ESMT F50-series QSPI NAND flash. flashrom doesn't support it, loading the spinand.ko module on Raspberry Pi detects the chip correctly but fails to access its contents. My suspicion is the SPI controller, so I thought about synthesizing a Cadence QSPI into a Zynq-7000 and accessing it from Petalinux.
I have no idea how much effort it would be to write a SPI-NAND applet for Glasgow.

2024-11-11 16:47:28 UTC

- reply

Exactly.

2024-11-11 16:46:26 UTC

- reply

Interesting, I would've put a second piece of stripboard, 90° degree rotated on top of it.

2024-10-19 10:49:50 UTC

- reply

Found it: http://ftec.cfasp.de/bre/spezial-schrauben.htm

2024-10-19 09:38:05 UTC

- reply

Finally someone took the effort to machine them. I first saw them as drawings >20 years ago.

2024-10-17 05:51:53 UTC

- reply

This is a shitpost, right?

2024-09-02 16:47:14 UTC

- reply

I guess it's just missing the connection between the zener's cathode and the VCC pin.