SDG

[@steverobbins4872]

This layout may be pretty noisy. You should follow the rules of DC/DC converter layout: 1) Identify your current loops. There are two: one when the switch is closed; and the other when the switch is open. 2) Arrange your main components such that the loop areas are minimized, and (this is very important) all on the top surface layer. For example, DON'T use the ground plane to connect CIN and COUT together. 3) Use a "single point" ground, meaning that any ground vias are all clumped together, not spread out. Rules 2 and 3 go together because they prevent any large switching current from flowing in your ground plane, which causes common-mode noise in the sensitive parts of the board. For example, in your case the only sensitive point would probably be your linear regulator; any CM noise present in the ground plane where the linear regulator connects to ground will appear at the output of the regulator, which kind of defeats the purpose of the linear regulator. Look at how TI did the layout in the data sheet and follow the loops. First loop (switch closed) goes from CIN positive, through the FET (inside the chip), then through the inductor, then through COUT, and finally back to CIN via the ground. (This first loop is fairly tight, but TI could have made it a lot tighter with a 2-layer board) Second loop (switch open) starts at the diode cathode, goes through the inductor, then COUT, then back to the diode anode; very tight. Note that they used polygons instead of traces to minimize the open area inside the loops because it reduces EMI. BTW, another issue you might have is beat frequencies because you have three switchers connected to the same source with no chokes to separate them. I hope this was helpful. I don't mean to be critical.

[@sdgelectronics]

Thanks for your comprehensive comment. I actually took some time to review the layout after uploading the video so most of this was caught before ordering. I've left the ground plane however - there's mixed thoughts on this as it depends on the frequency of the switch mode controller. At >200kHz, we'll just be starting to get to the point where the return currents on the bottom ground plane will follow the path of the current carrying trace on the top side of the PCB. Hopefully we can perform some EMC analysis at a future point.

[@rjordans]

Another thing that I found had quite an impact is the capacitance of the switching node (connecting output of the IC to the inductor) to ground. A few picofarads there and the spikes you see on the output get quite big, scope probes wreak havoc there. Anyway, avoid using a large polygon there!

[@steverobbins4872]

@@rjordans Or, you can use a polygon for that node, but just don't have any ground plane under it. Also, often the reason you see big switching spikes is because of resonance. It's typically the parasitic cap you mentioned, plus the parasitic cap of the inductor in parallel with the parasitic inductance of the output caps. You can often significantly reduce the peaks of these spikes with a series RC network in the right spot; the C is large enough such that its impedance is much less than R at the resonant freq. And the R is typically around 10 Ohms to provide damping to reduce the Q of the resonator. I'm not sure I know a good design procedure for working these out -- I don't do many power supply designs these days -- I usually just do trial and error. But I think there are design equations out there somewhere.

[@rjordans]

@@steverobbins4872 Thanks for the hint. I was wondering indeed what a snubber would do here and how to put one in. A quick search got me to a TI technote "Minimizing Ringing at the Switch Node of a Boost Converter". It's for a boos instead of the buck regulator that I was looking at but at least gives some guidance to start experimenting.

[@steverobbins4872]

@@rjordans Thanks for the reference. This looks like a good article. However, for the record the author (like most engineers I've met) may be misusing the term "snubber". According to my text books a snubber is an RCD (D for diode) circuit that reduces the stress on the switch by altering the trajectory of its Q-point; it is not intended or designed to reduce ringing. I think the RC circuit should more precisely be called a damper, but I haven't found any references to support that.

[@anonimuso]

This is an amazingly informative video for a computer engineering student like myself. Thanks for taking the time to do this.

[@rogerfurer2273]

Thank you so much for this video. I have not done a pcb since the early 90s when I used to etch my own after doing the layout with a simple CAD program plotted on transparency. It is great to see a step-by-step guide to the whole process. It improves my confidence level immensely.

[@cassio-eskelsen]

This was a complete class! This video is long but the way you explain it doesn't make it tiring. I like all your videos but this was one of the best so far :)

[@DustinWill]

I always learn a lot of tips-n-tricks from this style of video. Keep it up.

[@steveng5503]

Brilliant! At last I found a purpose for KZbin. I followed you intently from the outset and understood you for the most part. Looking forward to the next episode. A BIG 👍 from me! 🇬🇧

[@zopilotesky]

Very informative indeed. Just the parts selection portion was worth the price of admission. 👍

[@meatgoat4084]

Brilliant episode. I can honestly say I didn't want it to end.

[@ZeedijkMike]

Pure pleasure to see your work process and the result of good preparation.

[@chefbink61]

I really enjoyed this vid I got quite a bit out of it and you really do a great job explaining your thought process. thanks for sharing look forward to more vids like this.

[@akhurash]

Great video. When designing SMPS I like to put in couple of small value ceramics right at the input of the module and at the output of the converter. This helps to reduce high frequency noises. It’s not always necessary but can help to pass EMC requirements.

[@bloguetronica]

I always had the temptation to run away from the Simple Switcher DC-DC converters from TI, just to avoid the use of a external freewheeling diode. However, the phase margin compensation feature makes them a suitable candidate for future designs. As for the capacitors used, I tend to choose the expensive solid/polymer electrolytic types for such applications. For other applications that don't require high voltage and for relatively low ripple applications (for example, local decoupling, along with ceramics), I tend to choose those from the OxiCap NOJ series from AVX. Those are niobium types, and are much more tolerant and less explosive that tantalum caps. Anyway, it was great to see your design process. Thanks for sharing!

[@DennisGentry]

Late to the party, but this is a terrific walkthrough of the design process. Besides the spectacular failure modes, I try to avoid tantalum caps because tantalum is a "conflict mineral" mostly mined from the Democratic Republic of the Congo, not typically ethically.

[@IanScottJohnston]

Really good theoretical video but very practical at the same time.....nice balance. You'll get a nice psu you can use elsewhere also.

[@rene-jeanmercier6517]

Thank you for sharing your design approach. Very informative and instructive. Looking forward to next step. Regards, RJM

[@Adrian_Finn]

Yes! been waiting for this, what an excellent and informative video and i learned so much about the design process. Thank you very much for sharing and i will definitely build this for my FY6900, looking forward to the gerbers!

[@brucefox6580]

Thanks Steve, outstanding effort. Will come in handy. 🇦🇺

[@power-max]

Didn't Dave Jones make a video on how linear regs aren't ideal as active filters and noise suppression? He showcased a neat "capacitance multiplier" which was really just a buffered RC filter.

[@CoolNamesAllTaken]

Fantastic tutorial video as always! Looking forward to hearing about creepage distances and high-voltage considerations.

[@deangreenhough3479]

Great work🏴󠁧󠁢󠁥󠁮󠁧󠁿

[@mariushmedias]

I would have looked at AP63201 for step-down (AP63201WU-7DICT-ND on Digikey) ... cheap, high frequency so cheaper surface mount inductors and capacitors. Very detailed datasheet with recommended values for components AND suggested circuit layout. 1.2$ in 10qty. AP63200 is same but in TSOT26 footprint. Another option ... Richtek RT8279GSP , 1.3$ if you get 10 (edit: it's 0.7$ each on LCSC, 9k+ parts in stock), 5A 500kHz and then you can power the inverting regulator from this rail, along with the 5v regulator.

[@PapasDino]

Thanks for the design process video Steve...learned a lot!

[@irishFiddle0]

Thank you Steve! I tried to upload the Gerber files from your website to JLCPCB quote page and after the upload and navigating to the Gerber viewer it says "The new Gerber Viewer doesn't support files upload before July, please re-upload on our quote page to try the new version.". Any chance to reupload updated files compatible with current JLCPCB web? I'm new to electronics in general so I apologize for possible error on my side.

[@rjordans]

Nice tutorial. I'd probably try to keep the 24V input pour restricted to the input sides of the converters though.

[@Dumont67]

Nice video m8, good narration and much information

[@robertcalkjr.8325]

I must have missed something - so you are saying that the polarity of the voltage of a FET in the electrical characteristics in the datasheets are meaningless?

[@robertcalkjr.8325]

Never mind, I found it.

[@Electrolab28Ag]

Hi SDG, I just bought the FY6900 Signal Generator and I watch your video on the power supply modification. I think it is a brilliant idea and would like to adapt that into mine. How do I get the PCB with the parts?

[@ivanreynoso3849]

Please do more design videos!

[@grahambambrook313]

I am a mechanical engineer (retired) and in a former life in the mid-eighties, I used to design PCB and thick film hybrid layouts for a living. Looking at various videos like yours on YT I now wonder how the hell we ever managed without CAD. I note you did not mention the auto-router in your package; did you ignore it in your layout? Manual layout is undoubtedly a much longer process but offers a real sense of achievement when you 'get it right'. I'm sure I still have my scalpel around somewhere. :-) Also, I would love to know why you use electrolytics and ceramics in parallel. Is it a frequency thing?

[@KissAnalog]

Electrolytic capacitors are not great at high frequencies and they have relatively high esr (equivalent series resistance), but they are there for low frequency bulk capacitance. The ceramics are great for high frequency and have low esr, so they provide lower ripple voltage at the switching frequency.

[@grahambambrook313]

@@KissAnalog Cheers! I thought it must be something like that.

[@tablatronix]

What about Niobium Oxide caps as alternatives to the tantalum ?

[@Robert-un3cf]

Webench is a nice tool. I'm a big fan of TI.

[@tubeDude48]

QUESTION: Why not use a transformer 15-0-15, with regulators for the LM317 & LM337?

[@sdgelectronics]

This is basically what I did for the older FY3224S/FY3200S, however there is a fair amount of heat generated by the transformers and regulators when driving 50 ohm loads. The DC-DC converters get closer to the design voltage without dissipating so much heat.

[@1959Berre]

I really enjoyed this.

[@cookieboi4449]

I haven't a clue about what I am talking about. But why not use a $25 pc computer power supply. You can get 24, +12, -12, +5, -5 volt. Drill a hole in the FY6900 and insert a 5 pin socket for a 5 pin plug. Now you have an external power supply for the signal generator. What is wrong with the one that came with the unit? Or is this just a lesson in electrronics engineering? Good video desription of what you are doing. I have a FY6900 and I am not going to change the installed power supply, but if I am then for 30 bucks I can have an external supply.

[@DoItYourselfMusician]

Very good video. I'm often completely lost when selecting components. What is generally considered a low ripple current? I always hear "low ripple current" mentioned but what value is that? I'm sure its dependent on what you are designing but is there a rule of thumb? Also what PCB software are you using?

[@MartenElectric]

Thanks for all the details Steve, are you going to post schematics and gerber files on your blog?

[@sdgelectronics]

Yes, and the bill of materials :)

[@MartenElectric]

@@sdgelectronics Now I'm not sure if I my sensors are overwhelmed by sarcasm, or you meant it, but thanks for a great video!

[@sdgelectronics]

@@MartenElectric I'll be putting everything on my blog so anyone can replicate the build. I know a lot of people were interested in upgrading the power supply section

[@MartenElectric]

@@sdgelectronics Indeed, thanks Steve

[@joopterwijn]

Did you not need also -5Volt?

[@josemesquita603]

Very good, thank you!

[@bgdwiepp]

Purchasing from LCSC, aluminium polymer capacitors are very cheap. These are quite low ESR and ESL and for the switching frequency of the simple switcher part should be sufficient enough to not use the ceramics.

[@eldenchang]

Nice video What EDA design software and layout are you using?

[@rahulkushwaha1984]

Why don't you use easyeda?

[@andymouse]

Its great, longer ?...no problem ! cheers.

[@puddingfrais]

Hello! I want to make it, as one of my first project. I want to order components, but how many watt are the resistors? Thanks

[@sdgelectronics]

They can be 1/8W - none of the small resistors are dissipating much power

[@dudleywin7514]

When you say you put a 50ohm terminator on.... at 3:30... is that a load of some kind?

[@stevec5000]

Yes, he put a 50 ohm load on the signal output to see how much power it would use that way.

[@dudleywin7514]

@@stevec5000 Thanks for replying Steve. I had to ask the short version of the question to get an initial response. So here is the long version... Terminology is the basis... I presume a terminator is a connector with a resistor in it. This resistor allows current to flow. If it was an inductor as a load then would the load cause a difference to the signal generator behaviour? So are there different loads that cause the signal generator a problem?

[@stevec5000]

@@dudleywin7514 Right, a terminator is just a connector with a 50 ohm resistor in it. Carbon resistors are generally used in low power applications since they are not inductive. An inductor isn't likely to cause any problem for the signal generator but it might be resonant at different frequencies which is not what you want when taking measurements.

[@dudleywin7514]

@@stevec5000 Thanks a lot! So I could say the carbon resistor is the perfect load because it has the right resistance and absorbs the current provided/pushed/drawn by/from the generator and turned into heat.... ?

[@jonathanfulcher602]

Wow, this is so informative! When do I use tantalum and when ceramic capacitors in switch-mode converters.

[@1959Berre]

Avoid tantalium caps. When they fail, they short with a vengeance, causing lots of troubles to whatever lies beyond. They can ruin a complete power stage.

[@parapos]

finally , thank you .

[@romancharak3675]

Very interesting and education! Thank you very much.

[@Sam-jw1yn]

Good video! I would like to know what CAD software you use. Keep up the great work!

[@sdgelectronics]

It's Labcenter Electronics Proteus

[@Mr.Leeroy]

41:50 right here is probably a good idea to remember about output offset feature of generator if it has any

[@rfmonkey4942]

very enjoyable.

[@laboratoriodeprojetos1351]

Ótimo vídeo Steve. Abraço

[@xDR1TeK]

And I thought I have to go apeshit on design because as an engineer never had a course in Uni about design stage. This is hardcore down to earth course. Thanks. Should I go burn my books now? I'm pretty sure some things would be out of bounds. Would be interesting to see how you adjust things around at the second run.

[@alejandroperez5368]

LM2676, currently on LCSC: 0 in stock :/

[@aherontas1943]

Very nice video keep up the good work! ;)

[@robertcalkjr.8325]

Thanks for the nice tutorial, Steve!

[@Chupacabras222]

Isn't it suitable to put LC (PI) filter between switcher and linear regulator? Linear regulator (like LM317 and LM337) are very bad at filtering high frequencies which are generated by switcher. Or it does not make much difference?

[@ЕвгенийСтародубцев-и9о]

Steve, greetings from Russia! It's a very good job. Did you save the Gerber files for the PCB? Can I buy them with your SDG authorship or buy a ready-made Board without installing components (or with installing components) through PayPal? I also bought Such a feeltek FY-6900 generator and would like to replace the Power Board. Success in business!

[@sdgelectronics]

The files are on my website: sdgelectronics.co.uk/fy6900-signal-generator-power-supply-modifications/ feel free to use them to get your own PCBs made.

[@stevec5000]

I know the purpose of this video is to show how to design a power supply but for practical purposes why not just pick up a readily available commercially made +5, +/- 15 volt power supply and stick it in the FY6900?

[@ProgrammerEA]

Good job! LIKE 👍

[@etmoimeme]

Great video, thanks! I like octopart.com to check parts availability (findchips.com might also work) (checks if farnell, RS, Digikey have the parts in stock.)

[@robertssmorgasbord]

Hi there, I did some rather amateurish measurements of the output stage performance of an unmodified FY6800 ( kzbin.info/www/bejne/q2fdgoBpiJpobKM ). In the comments to that video the question arose if with a FY6900 with your redesigned power supply is actually able to drive a 50 Ohm load at 20Vpp set voltage (an unmodified FY6800 can't - output voltage breaks down / waveforms are distorted above 10Vpp set voltage with a 50Ohm load).

[@RomaineRC]

I really wish this was edited

[@tylerufen]

So attractive, it's a shame your employer doesn't tap your phone line... Location, maybe?

[@justcallme00oogy]

You weirdo! He's all mine!

[@powersupplyguy3894]

love the videos but am I the only one that thinks he looks like the angry video game nerd lol its the angry electronics nerd

[@bfx8185]

keep in mind nobody is interesting to your face :) may be for schematics and so :D my hint ;)