Doing some holiday shopping? Please check here for shipping deadlines to make sure your order arrives in time.

He's Back! "According to Pete" Takes On Differential Pairs

You asked for it (a lot) - so he's back!

Favorited Favorite 0

You might remember our video series "According to Pete" -- the series where our Director of Engineering Pete Dokter tackles common engineering and electronics topics. The last episode we filmed was almost a year ago! Pete had a lot on his plate, so we suspended filming for a while -- we always hoped we could bring it back someday, but we had to wait for the right time. Well, the time is right and today we are excited to announce that "According to Pete" is officially back!

In today's episode, Pete is diving into the world of differential pairs. What does that mean? Watch the video, wouldya?!

As always, if you have any questions, leave in the comments section below. Hope you enjoyed this episode of ATP!

Comments 24 comments

  • Hi Pete. I am very glad you found time for the videos again. I really enjoy watching them. Keep up the good work. You are fantastic.

  • So is this what USB uses in the D-/D+ differential pairs?

  • Don't have as much time for HW hacking as I'd like, but am a bit of a cerebral guy. I much like these and had just started following them when you stopped. I'm glad you started again. They're about perfect IMO. Deep enough to be intellectually challenging on (more than just basic E=IR anyway) but not so deep that I'm forced to commit hours of brain time to figure out what's going on. Definitely thanks for restarting these!

  • It might be usefull to include some circuit simulation.

  • Awsome video, forgot what Vt was but I remebered it was 25mv but forgot where it came from, lol. Took me back to Semi II in college :)

  • Hey everybody, thanks for all the kind words. For the next one, I'm sorta thinking I want to play with some oscillators... will will tick Gregg off cuz it's another huge topic. That, and maybe dropping about 10 pounds and getting a haircut.

    • Oscillators would be great Pete. How about a Pierce oscillator, show how to drive a crystal? Or maybe the classic Wien bridge oscillator. Make a fast one in the MHz range and you can talk about op amp selection to boot (slew rate, etc).

      • Ah, sweet-- the simple 32MHz discrete setup in this M1 is called Pierce. Thanks for the clue. I could wish it was as easy to find names for things from descriptions as it is to find descriptions from names, but now I don't even have to hunt. +1 for oscillators because the differential pair sort of reminded me of Royer oscillators as in CCFL inverters.

    • great! you should do a colpitts with a xtal in series resonance mode in the feedback loop, very stable and easy.

    • I like your hair the way it is :)

  • Welcome back, we really missed you! I'm looking forward to watching the video, but for some reason it won't play on an iPad. This started around Thanksgiving, and now even videos that would play in the past won't play anymore. :( Of course, the other videos weren't important enough to get me to actually comment on them!

    Guess I'm going to have to wait until I get to the real computer...

  • Hi Pete, Had an encoder directly coupled to a stepper motor but had trouble reading it above ~10 rpm. Encoder signal got too fast and noisy. Found a differential line driver to boost the signal and cancel noise via common mode rejection and now it keeps up just fine. Wasn't quite sure of what I was doing but after watching your differential pairs blog, at least now I understand what I did! Thanks.

  • Thanks for putting the subject in the video title. When I hit a bump, (or a part jumps off the board) I can search your posts and recover my nerves sooner. P.S. the differential pairs is spot on for a current communications project. THANKS!!!

  • Glad to have you back!

  • Up the dosage! (lol)

  • Hi Pete, I think it would have been easier to follow if you'd have started with a long tailed pair made of "ideal"-high-beta transistors, degenerated with emitter resistors, so that the gain clearly is Re/Rc... If you start out with the configuration without emitter resistors, your circuit starts out as highly non-linear, hence for a practical application, you'd want to introduce some feedback (the long-tailed pair is often used for "discrete opamps" in audio, anyway) to avoid distortion (unless, of course, you build it just for distorting your guitar...), or have very, very low input signals.

    Here's small example in partsim where one can play around with the different values of resistors...



    • Thanks for sharing your sims! Your upper transistors are very close to saturation - you won't get much output swing, but you will enjoy much distortion (grin).

      Better to change the bias to make the common-mode output voltage closer to the mid-point (say, 7.5V in this example). For the 8k2 collector loads that requires about 300uA in each side. The 'tail current' should therefore be 600uA not 1.17mA as show. I'm not a PartSIM user (prefer PSICE), can you run a fresh sim to confirm?

      • ...guess what... I wanted the example to show some nice distortion (see 2nd image)... ;-).

        If you click the "Spice/(>)Run" icon, you'll find a spice input file in the right half of the "Report" tab. Or just change the parts in-place ;-).

        • Thanks! If nothing else you've got me interested in using the partsim tool. BTW, 600uA gives a VCM output of 7.55V

Related Posts

Recent Posts


All Tags