Circuit design

Thread for posting and discussing circuits you're working on, have worked on or just think are plain cool/useful. Stuff like: circuit bending, RF, audio amplifiers, microcontroller based circuits, circuits for bigger projects you're working on, unusual circuit you found, etc. I've been working on an ESR meters for capacitors. I was inspired by this 5 transistor design: https://www.eevblog.com/forum/projects/5-transistor-esr-meter-design/ . I've gotten my design down to 3 transistors, looking to get down to if I can remove the buffer or amplifier stage. Transistors are cheap as dirt but I find it cool to optimize for simplicity, you'll noticed I managed to replace the original differential amplifier Wheatstone bridge setup by simply using an amplifier and a peak detector! (Output is read by a Voltmeter instead of an Ammeter like in the 5 transistor design). The oscillator is a "simple" joule thief style blocking oscillator (simple in components but very complex in operation), merely using the LED in style of a zener diode (a couple of normal diodes do the trick too). The transformer for it will probably have to be made by hand as with most of these small blocking oscillators. It operates at around 100kHz, which is good enough to measure capacitors of around 0.5uF and above, any lower and the reactance at that frequency will be measured as a false ESR. This should be fine as ESR is only really a concern with biggish power supply caps which are always above that. Oh, the voltage on the cap is always lower than about 0.15V, which should make it possible to use in circuit, removing the need to desolder the caps to test them afaik! Here's the qucs schematic for it: https://pastebin.com/kMEDDuSf

Bumping with yo boi, blocking oscillator, back again. This time it's used to drive a center tapped transformer (driving from the middle tapped side, reverse of usual) to generate around 4kV to 8kV: https://www.youtube.com/watch?v=aR0y4xnS6h4. Obviously, you could theoretically get any voltage you'd like at the output, but more than a few kV will be hard to maintain due to losses and require more significant isolation in the transformer. At these high output voltages it's already very unstable and basically unable to drive a constant load. Still good for half centimeter spark gaps tho! These are sold as "arc igniters", which they are surprisingly well suited for. I've attached a better schematic of the circuit, please do note that the transistor will see 100V or more on its collector, so a generic transistor (like the BC337 there) will quickly release its magic smoke. You'll want to use a very resilient power transistor if you want it to last. A high voltage Zener across the collector-emitter of the transistor would really be the best solution tho, to clamp down that voltage to acceptable levels for the transistor.

Would this be a good thread for me to post pictures of my Dad's old receiver that I think might have some blown capacitors? I'd like to rebuild it, all the components are large and should be easy to replace

Hey y'all I needed to make a circuit to use a photodiode BPW34. I started with a lowpass filter but I dunno how to make the frequency response plot (bode). Is there really no way to sweep the freq? Like .step param?

>>7880 whats the program and OS here? I've been using kicad on linux but can't get the SPICE sim soykaf to work

>>8060 Not OP but it looks like LTSpice. ime it works perfectly with wine

>>8061 thanks lainon ill check it out

>>7880 One of the LTSpice pitfalls is that "m" equals 'e-3' ! If you want 'e6', you have to write "meg". Also, what you want for your diode is a transimpedance amplifier - which means the input should also be a current. So put at least resistor in series... Good luck, you are not too far off!

>>8065 use a capital M for that?

>>8455 You might care about capitalization, LTSpice does not. There's a reason it says pitfalls.

>>7880 I can't recall how I used to specify the desired test frequencies on the voltage source but i know it's possible since i used to do this on college, what I like the most of ltspice is that you can plot whatever you want, this may help you https://www.analog.com/en/technical-articles/how-to-generate-a-bode-plot-with-ltspice.html look into the ltspice manual, i'm sure somewhere it mentions how to setup the source for frequency swapping, there are also some resources on universities sites describing how to achieve it, i think it needs to be a different type of source, an AC one. kinda miss those times playing around in the sim and then testing the circuits in the lab, wish i had real use for a scope and and a freq analyzer, but stuff is too expensive and I don't build circuits anymore...

>>9615 Modern e.g. Siglent scopes have the generator built in and will do the Bode plot for you quite happily, out of the box. Now sure, ~1k *is* pretty expensive. But compared to before times, it still is pretty affordable (and a versatile tool in many other directions, like serial protocol decoding). Also pretty small, light and low power compared to old iron HP. OTOH, many thing can be accomplished already with simulation, a decent multimeter and an makeshift "audio response analyzer" that consists mainly of a usb soundcard with line-in.

>>9617 I used to work for a calibration lab who was a collaboration partner of Siglent and we hated the brand. Siglent is the most rank of the Chinese brands. Barely above Hantek. Like, it's literally just Hantek tier equiment that they made conform to some industry standards. For some damn reason we never got any Siglent scopes to try out but lemme tell you that the specifications on their DMMs are lies. They spec them like the competing Keysight or Keithley DMMs except they don't hold up. Very high rate of being out of tolerance. Now, scopes arent really precision instuments given most have 8 bit ADCs and vertical gain specifications of +/-3%. But I'd avoid Signlet alltogether considering how rank the brand is. Go get a Rigol at least. I've seen a Rigol device being within tolerance for 8 years in a row. For Chinese brands that's outstanding.

>>9618 Interesting that their DMM's seem to be duds, but since they are an OEM for the low-tier Lecroy scopes and waveform generators, there should be a good chance that at least these are decent. Just out of intrest, was that on a specific model of DMM or in general?

>>9619 Their SDM3065X and SDM3055X. Look at the 1-year voltage specs of the SMD3065X for the 2 V range: +/- (0.0035 % reading + 0.0006 % fullscale) Total nonsense. It's easily tripple that. All technicians dreaded having to calibrate these because there was an additional inspection procedure for devices found to be out of tolerance which required the head of laboratory inspecting your work and it was basically every time with Siglent. Personally, I'm also not a big fan of Lecroy scopes up until their HDO6000 series. Frankly, the job kind of spoiled me as I used to have daily $20.000 - $40.000 scopes at my fingertips. But I can vouch for the low end Rigol scopes because I did a lot of those, too.

>>9621 The Lecroy user interface sure is *particular*, so I get the sentiment. The Siglent UI is different and IMHO pretty easy to "get into" even without a manual, which is usually the hallmark of good UI design. I've had my hands on the Rigols as well and to be honest I'd recommend the Siglent any day of the week with regard to ease of use. Now with regard to these drifty DMM's, did they by any chance share a common environmental denominator? Maybe some of the components used (SMD resistors come to mind, among other things) have a particular weakness to a thing like trace Sulfur or an trace corrosive gas (like they are used e.g. in a Fab) or maybe prolonged high humidity?

>>9627 Yeah, Rigols have absolutely soykafty UI, I can totally agree with that. I have no idea what the issue with the DMMs was. We didnt get any technical informations out of Siglent and we weren't allowed to open the units ourselves for repairs. Faulty units were simply shipped to Siglent for maintenance, we only did calibration and adjustment for them. Despite the collaboration contract, their non-Chinese technical devision was useless for us. I remember we contacted them a couple of times for clarification on how to get access to some settings or what some of the datasheet entries meant which we didnt understand and they replied "Franky, we don't know either. The Chinese division doesnt tell us anything. If you manage to ever figure it out yourselves, please pass it on to us." Yeah, nice, wow.

I want to get better at circuit design, currently using KiCAD since I can take what I make and export it to get fabricated, is that considered very good for circuit analysis or is there a better alternative on Linux?

>>10694 Make stuff and measure it sure is the best way to go! To test simple ideas, work out concepts or analyse specific problems, spice sure helps. LTspice works with WINE pretty well, qucs-s is linux native and can handle different solvers under one UI. You can go the hardcore linux route with the XYCE solver with Geda for schematic capture. But: to get "real" results with any spice flavor, you have to put the relevant parasitic RLC influences from real world layout and component non-idealities into the schematic as well. Think trace resistance, component lead and self inductance, parasitic capacitances between layers, leakage resistance from one trace to another etc., as applicable and relevant to the circuit and characteristic in question.

I have an EE degree but my circuit game is pretty weak due to having worked as a software developer for however many years now. I recently bought a vintage synth I had wanted for some time. Unfortunately, it had some garbled audio output. There's an AV repair shop in town that sometimes does 'fix your soykaf' meetups, so I took it over there. There was a girl who worked there who's absolutely brilliant and we spent the evening reading schematics and scoping things. Watching her diagnostic process was impressive, you could tell she's just been fixing audio equipment day and night for years. We didn't fix the synth, but it got me stoked about fixing and making things again. I've been building some open source hardware projects related to composite video generation and FX. My friends have been doing some markets around town and I setup a table with some examples once to see if it was something people were interested in. The response was good so I'm going to try selling some of these AV/glitch boards in a few weeks at the next one. This week I watched this guy Mortiz Klein on yt do a deepdive on designing and building a Voltage Controlled Oscillator for a modular synth setup. The video series is really excellent and I highly recommend checking it out if you have any interest in learning to design things w/ transistors.

Not strictly within thread specification, but as somebody with no formal EE experience hacking on an electronics project that just might make some money, here's a book I found obscenely helpful for demystifying some of the mystique of PCB design.