I have most often seen Sziklai pairs in the last stage in audio power amps, with a small-signal PNP as the input of the pair and a power NPN at the output. The combination behaves like a PNP but has the current rating of the NPN. This can be a cheaper way to get a power PNP transistor, or in the old days maybe the only way. Congrats on the patents! That's an impressive body of work.
Yep in the old days of the 1970s this is how we'd get class B or AB final stages going, using the classic 2N3055. A PNP complementary to it was not available (or rather very hard to get, the PN2955) so the stage was quasi-complementary: a Darlington with another medium-power NPN+2N3055, and a Sziklai with a medium-power PNP (much easier to find) and a 2N3055.
i was about to type that still if you making negative ground devixe this is een better, could be a substiture for germanium darlington with no bias change
Except that the inventor of this circuit, George Clifford Sziklai has a Hungarian last name and in this case 'SZ' should be pronounced like 'S' in the word 'see'.
@@SLeslie : That's exactly what I wrote: 'SZ' is pronounced like a sharp 'S'. Not sure where you're getting at... My FYI is indeed about the way Hungarians use and combine 'S', 'Z' and 'C' and what the results would be pronounced like in English.
@@bikkiikun Sorry, I probably misunderstood what "sharp S" should mean then. Although I believe I do know how Hungarians combine letters being a Hungarian myself
@@bikkiikun Am I mistaken if I believe you to be German? Does the "soft S" come from the German like pronunciation like Simson (the former moped manufacturer in Suhl) for you? Fun fact: Those mopeds were popular here and we definitely pronounce its name improperly the Hungarian way with sharp S. What I probably should have suggested from the beginning is if someone is debating whether he/she pronouncing a word right is to copy-paste it into google translate and read the word out loud with it.
The reason why the Vbe for the Sziklai pair is only 544mV is that the base current was 6.1mA/11,100 = 550nA. That's not an unusual voltage drop across a silicon p-n junction at such a tiny current. The Vbe's that you were expecting, 600mV - 700mV, are more typical of p-n junctions passing milliamps.
FYI a close approximation to Sziklai in general american is SEEK-la-yee, with the weight on the first syllable (but still short). The 'y' is a pretty good approximation of moving from the short 'a' to the short 'i'. It basically means "Of [a] boulder", which is what I wanted to bash my head into when I first started learning about transistors after basic circuit theory.
@@rich1051414 Hungarian doesnt merge vowels into full dipthongs, they're used as a transition when needed. 'Lie' omits the trailing 'i' which is a case suffix (the language is agglunative, so all cases are basically either prefixes or suffixes instead of seperate words). 'Lie'+'ey' might be a close approximation but its still a little off to my ears, mostly because the short 'i' is a very regional thing in English use IMO. But honestly it's splitting hairs, any attempt is a welcome attempt considering how little people are even exposed to the language's existence.
Big difference in the "on" voltage of the two configurations. The darlington cannot go lower than the EB voltage of the transistor while the skezke configuration can saturate to millivolts.
That's why he's retired ;) but most of his patents must have been filed at HP which owns the parents and gives out a little fixed payment when filed or granted. Also, some patents would be associated with more than 10 people.
in the US you get zero dollars. if you are lucky they might give you $100 or $1000 when they file it. in other countries I would be able to retire sooner, you get a good percentage. some US companies give percentages too, just not the ones I worked for.
@@IMSAIGuy Seriously impressive sir! 60 patents is incredible! 👏🏻 Additionally, your videos are very educational and I thank you for that. Keep up the great work.
@3:50 It's 0.544V instead of 0.6V BE voltage due to the very low base current. The diode characteristic is a e-function. If the current is lower than usual you move more to the left in the e-curve.
NPN silicon going to a Germanium PNP transistor Sziklai is a fun Sziklai for Germanium Fuzz guitar pedals. It greatly reduces the leak from the Germanium transistor in the pair.
The 'leak' would get more severe the harder the base NPN is pushed. Would that result in a dc bias when driven very hard? I guess it wouldn't matter if pass through a capacitor to remove any bias anyway.
In my experience, if the pair isn't too thermally sensitive and you can find a reasonable operating point, you can make a fuzz out of it that someone will find interesting! To that point- in college, in 1998 or so, I built some fuzz faces for friends with used germanium from a transistor radio, by tweaking the resistor values until they sounded great. One in particular sounded amazing. I didn't have RG keen's measurement rig yet, so my tester showed the input transistor had a gain over 200. I'm certain it was just leaky. Whatever the case, I managed to bias it up by ear and the results were great. I guess, in the end, there are no rules as long as it sounds good!
2:30, if the hFE of a transistor is 173, let's round it to 200, so putting 2 transistors in darlington pair configuration should give 200 * 200 = 40,000. Or roughly 20,000 to 60,000 not 862,000. What's going on?
I may be wrong, but I think a Sziklai would be more temperature stable than a Darlington. PNP and NPN transistor combinations are used to stabilize temperature drift of control voltage in synthesizers fairly often because they behave oppositely to an increase in temperature.
The Sziklai is usually implemented with a bias resistor on the second transistor and there are 2 topologies. One disadvantage of the Darlington is its low frequency response, compared to a common emitter stage. While an emitter follower stage, using a 2N3904 for example, can have a frequency response of 2 MHz, a Darlington stage will only operate into a few tens of kiloHertz. The Sziglai has a better frequency response than the Darlington,. I haven't done a calculation (sum of the inverse time constants) for the Sziglai so can't give a more precise answer. The cascode and differential amplifiers, which are both 2 transistor stages, have much higher frequency responses (~12 MHz) and higher gain than a common emitter amplifier.
I came up with this configuration on my own when trying to avoid the inability to saturate a Darlington pair unless you have a voltage at the first base 1.2V above the second transistor's emitter. It just seemed like a very simple complement to a Darlington, and handy when you wanted to fully enhance a PNP power transistor, but didn't want to expose the driving circuit to the potential high voltage on the PNP's base (ie logic).
Also the Sziklai pair is the basis of your IGBT power device, integrating a high voltage low gate capacitance n channel mosfet with an absolute brute of a PNP power transistor, and a diffused base emitter resistor, plus a massive body diode thrown in almost for free as well. Low drive needed for the mosfet, and it can turn on and off fast, and the power PNP transistor can suffer the miller capacitance without reflecting into the gate as a huge charge you need to either insert or remove. So the advantages of a mosfet of not needing huge charge to drive for a huge current, but also no huge gate capacitance either, even if the overall device is relatively slow in acting, but it does have very good pulse rating and also a very low forward voltage drop, unlike the darlington, which will always have around 1V5 of drop across it irrespective of gate current, due to the 2 base emitter junctions, and the first transistor needing to have some voltage across it, plus the second transistor will never saturate.
In 1958 IBM actually employed more engineers than the "Lazy L" as it was called at the time. IBM was developing the secret Winchester Drive at that time, replacing the drum drive.
Great discussion, glad to have this shown to the masses! Maybe some enterprising young engineers will consider the Szilklai in their own designs. I see this in a lot of AF/RF amps due to its ability to work at lower bias levels. I'm not sure how much of that is due to actual circuit knowledge or just copying libraried designs. An interesting topic that might be fun to cover is the difference (probably no more) between Japanese 'electron-flow' teaching and western 'voltage flow' - *and* how that influenced the design of amplifiers, increased the use of PNP transistors in Japan, and how it makes 1960s-1980s Japanese schematics so darn hard to figure out ;) Great video, love the content. Cheers!
03:52 that's because Vbe determines the Ic of a transistor, and in this case the collector current for the NPN tranistor would be reaaaally small (since it's the Ib for the PNP that has a current of 6mA trough it). Even in a darlington the Vbe of the driving transistor will be smaller than that of the power transistor.
The Sziklai design is very convenient for amplifier power output stages. At first you don't lose less output voltage swing, because it can be driven closer to the + and - power rails. At second the temperature compensation of the idle current of the output stage is less critical because the driver transistor gets less hot if you oversize it a little bit (use a 500 mA to 1 A device if you have to handle 50 to 100 mA) and don't locate it in the direct vicinity of the output transistors. It is also good practice to use feedback resistors in the Darlington or Sziklai designs to limit the total current amplification to a reasonable amount (1000 to 2000) in order to get a more linear characteristic of the emitter follower output stage.
Built one of those IMSAI 8080 computers in 1976. Built 8 4KB memory cards for a whopping 32KB. We bought a Z80 board when those became available and on a good day you could flip the switch to 4MHz!
So if your aim is for power, I imagine the two transistors don't have to be identical or complementary, I think perhaps the "gate driver" transistor can be a signal transistor that's driving a power transistor.
Since in both configurations the resulting Beta is approximately the product of both transistor Beta values it can be useful to pick a large beta one for the smaller one, as it needn't be rated for the output current, nor heat dissipation ability
At 1:09 you have pronounced his name pretty mutch correctly, it's ' Sikla'y ' . The 'sz' in Hungary is meant to create the 's' sound while the vocal letters do not merge in 1 sound.
The times I've seen Sziklai pairs being used is in power amplifiers before silicon transistors. It was significantly easier to make germanium PNP power transistors than NPN ones and with a Sziklai pair you could omit the use of hard to get and expensive power NPNs :-).
I also believe that the darlington is slower. There is a capacitance Cbe that can charge quickly if the base voltage rises but then you don’t have a good way to discharge it on the way down.
Interesting. I've been wreaking electronics since the 1970s, and this is actually the first time I've ever heard of Sziklai pairs. They're intriguing, but not overly handy.
Here is my observation... - Sziklai is pronounced differently in USA and the rest of the word. When I moved to USA I had to relearn to pronounce a lot of words such as Boston which is pronounced "Baas ten" here. - Here is shown an NPN driving a PNP (power) transistor. Power silicon PNP were harder to make and were much more expensive. For example MJ2955 was many time more expensive than its NPN counter part 2N3055. The Szklai pair is counter to reality in silicon age. But one thing we forget is Sziklai came up with this combination in germanium eara. When he invented this circuit in 50s, it was after transistors had been invented and they were mostly made with germanium. In that eara, PNP germanium power transistor were more common than NPN. I have never seen an NPN power transistor myself, and I have seen a very few low power NPN only. That's why you always see an NPN driving a PNP transistor when someone describes the pair. - Both of these pairs have their issues like the frequency responses/the unity gains frequencies, and saturation of driving transistor. That's why you only see them at the power output stages of audio amplifiers. Another more common circuit is using both in common emitter and driving the base of power transistor with a resistor. That resisitor limits the max current to the power transistor. - I'm trying to earn 3x1 cents here ;)
It's very interesting. In Hungary the used name of "Sziklai" is composite switched transistors or simply composite transistor. We never use the name "Sziklai".
To be fair, I'm not all that sure a simplistic "multitester" like that, with next to no analog front end at all to work with can be expected to resolve a hfe like that beyond "oh shit it's really high".
Thanks for the video. The video reminded me of the resistance measurement current source used in the HP34401A DVM. For the life of me I can't figure out how this circuit works. This circuit would make a great video episode. Part of the circuit contains two strings of 4 PNP transistors connected in series, an op amp and four selected resistors to set the current. Craig
To help a little - a number of European languages have letter combinations that seem horrendous in English! e.g.Polish, SZ is usually pronounced as SH (Shh, Quiet) in English. CZ is usually pronounced as CH (Chalk) The letter combination SZCZ (Oh wow!) is not too bad if you think of an English word 'Ashchurch' A suburb, - not too bad, eh? So, Can I suggest Sziklai could be pronounced as Shik-lai. Hope this helps. Best regards. Excellent lecture, by the way.
An interesting topic at some point in time would be some of your patents. Maybe make up a list of them so we can pull them up on Google Patents. Very interesting indeed.
I created an abstract from the transcript using Gemini Advanced 1.0: *Abstract* This video explores the configurations and applications of two transistor types: the Darlington and the Sziklai (pronounced 'Sik-lie'). The speaker begins by emphasizing the similarities between the two, noting they both serve to amplify current. The primary difference lies in their base-to-emitter voltage (Vbe). A Darlington configuration will have approximately double the Vbe of a single transistor due to the cascading of two transistors. Due to their lower Vbe, Sziklai transistors can be advantageous when used with low-voltage microcontrollers. For higher current applications, the speaker notes that NPN transistors remain more readily available and affordable, explaining the continued popularity of Darlington pairs. *Keywords:* Darlington Transistor, Sziklai Transistor, Vbe, Current Amplification, Microcontrollers
0:25 szikla=rock +i mark of origin. like newyork-i mean somebody form newyork. used as surname at medieval ages as house name to show where is the "headquarters" of the given noble family. but the szikla is not a city... so its probably a hungarised name form late 1800's (probably shaul) and not follow the naming conventions.
@2:13 Funny how you still managed to screw up the value even with the correction. 😆 Interesting video nonetheless, had not heard of a Sziklai Transistor before.
Interesting topic. The first transistor has a much lower base current and as such also a low Vbe which is a nice feature. I’m thinking what would be the behavior over frequency as the miller caps may work differently in the Szillai pair. I need to check this in LTSpice. Any thoughts anyone?
I would think the Sziklai would be able to de-saturate faster as a small resistor could be added in the middle to pull the smaller transistor out of saturation.
I will appreciate if you explain the functioning of a TL431 connected to the ground of a 7805 regulator to get a variable supply from 1 to 5 volts .The TL431 refrence pin is conected to a voltage divider between the 7805 output and ground .Thank you very much
that won't do. In such a circuit the voltage at the 7805 ground pin will be above ground. the 7805 output voltage wil be 5 volts above it's ground pin. However the 431 will sink current at the cathode until it's ref input drops below 2.5V. To regulate to 2.5V at ref input the 7805 output must be at 2.5 volts times the resistor divider ratio, which is achieved when the 7805 ground pin is at 5 volts below that. Within the 7805 some current flows from the input pin to the ground pin, which pulls the cathode of the 741 until the output voltage raises to the level that results in the 2.5V at the reference pin of the 741
I know the Peak Atlas will identify Darlington transistors... and I believe that some of the other low cost (probably Chinese) component testers identify Darlintons too. (... although it is probably just a flag in the firmware, when the current gain goes beyond a certain value... say 300?)
You forgot to mention that the Sziklai (my guess tsiklaiy pronunciation) configuration also has a much lower output saturation voltage, as it's about (1 x Vbe) + (1 x Vsat) where as for the Darlington it is roughly (2 x Vbe) + (1 x Vsat).
I learned the Sziklai circuit as the "White Darlington". I have no idea who What was. If anyone does, please let me know. BTW: Mr Sziklai's circuit can be done 3 transistors deep. A PNP driving an NPN driving a PNP can give you a really high HFE.
@@seanoconnor1984With high value emitter to base resistors, they also work to control heavy loads from low power circuits. I did the 3 transistor thing in a power supply I made for hobby stuff. On PNP drove 2 NPNs which drove 4 PNPs. It spread the heat out enough to let me get all the heat into a big heat sink.
I studied electronics for years ( high school ) but i never heard about Sziklai ( Sz is just S for english ) ... or i just dont remember this guy from the books !
I just tried signing on to you Patron to join your account there, but patron poeple are making it difficult. Interested in the tool you used to ID the chips. I am sure I can find it elsewhere (Walmart, Amazon...), but I don't want to give my money tho those things, I wanted to purchase it from you. ...
That`s the easiest part of Hungarian:) SZ it is one letter in alphabet two character. It stands for S in English. SIKLAI (S in Hungarian stands for SH in EngliSH)
Hey there! Great video! I haven't heard of the Sziklai pair before. If your trying to pronounce Sziklai better the 'Sz' written in hungarian makes the regular 's' english sound, the 'i' is like in 'hit', the 'k' is nothing special just like in 'kid', again the 'l' is just a regular l like in 'literature', the 'a' is pronounced like in 'around', and the last 'i' is pronounced just like the first one. Or, y'know, you could just use the google translator instead of reading my blabbering... Btw if someone is still reading this 'Szikla' in hungarian means big rock or boulder and it is also a name of a village. If you add an 'i' to the end of the word you'll get 'Sziklai' which means 'from Szikla' just like the name of Ludwig van Beethoven or Leonardo da Vinci.
Didn't know about this topology, I think this topology is being used in the voltage regulator circuit of my NAD 3140 amplifier... but I couldn't really make sense (not an electronics guru) of this part of the circuit, so might make more sense now... not sure why they didn't just use NPN/PNP darlington pairs as the circuit is dropping 10+ volts so 1.2 vs 0.6 vbe wouldn't have mattered, not sure why a voltage regulator would need the high gain of these topologies either 🤔
Both these configurations are great, just beware if they are used somewhere where they can "see" RF signals, they are a pain in the behind to stabilise. Don't ask how I know 😑
@@seanoconnor1984 Exactly! And emitter followers can be biased as to seem inductively reacting. I used that quite often to taylor freq. responses of circuits, but any inaccuracy in the extraction of the parasitic caps of the layout means a few hours in the lab debugging the circuit ... As a rule of thumb: an amplifier oscillates, and an oscillator amplifies, thanks Murphy :-)
Nobody can pronounce Hungarian names. It's impossible. 🙂 While I have long known about Darlington transistors, this is the first time I've heard of the Sziklai. BTW, I used to have an IMSAI 8080. I bought the kit in Nov. 1976. I remember people asking what would I use a computer for. 🙂
Thanks alot.... I knew you are a genius, a great inventor and HP was a good place where they encouraged IP... You have 57 more than I. At least we got like $2K each & better than a kick. We had a few weirdos...then liked to patent strange things...just for the money...but had little practical purposes.....I mean I admired their creativity... I am not even in that league of brilliant inventors... There was popular little Darlington TO-92 we buy by in the 1000's...plastic I have to look it up.. I think called an A10...Motorola Eariler during your Tek scope...and you gave up on the semi discrete design and replaced with 7805... I know I babbled...later it looked like there was another voltage...for I thought just 11 VDC dropped to 5....then I heard 19 VDC... In this Tek little circuit... It seemed to me..the comparator..if like a LM339..wouldn't it just switch on or off...and be noisy and output chopped then need filtered...so the comparator in the Tek... isn't it really a linear bias little Op amp....?..not a noisy switcher I like a better recording instead of my tiny phone...at work big display....no time to look up Tek scope +5 VDC or watch your TH-cam lession ... Probably head to Verizon...for a tablet... Our like fast curve tracer or computer that basically stored every bias point in digital...and list pass fail ....we had some Tek carts too just like yours on carts for nice entire curve display....Our main transistor tester was a Lorlin driven by the PDP-11...about the size of a small closet....plus the transistor handlers.... I mistakenly I called it a Loran...which ia a navigation system.... Our little AA community college....for those circuits 1986 ish we had the Malvino book...no calculus needed -I could hang with that...sort of simple and appropriate for the class I seen the Sziklai pair with the Darlington.....in audio amps...National Semiconductor great little old school AB bias...nice application notes....would show like what Malvino touted... NPN Darlington for top pair...and Sziklai pair for like the bottom or when the signal goes negative....the advantage being the Darlington and Sziklai when paried they both used both NPN in final pass with 3 diode....so crossover maybe less Malvino annoyed me somewhat...no real circuits from the standpoint of use a 2N5179...I like cracking it open...time to time... And NPN Darlington on top with PNP Darlington on bottom...is usual...but maybe tougher to find NPN PNP complementary pairs...plus there were 4 diodes...to help bias... So cross over distortion must be worse when your final audio amp uses NPN Darlington on top and PNP Darlington pair on bottom.....4 compensation diodes I thought he was Japanese too! Thanks for the enlightenment of his accomplishments and reminding us of yours... Another reader asked a few of say your biggest most successful..but then you might be worried we might look up your name... When I worked at IBM...to install and test semiconductor tools...they had great patent reward system... Luckily...I never tasked to build...soild state audio amp...which I like to do if assigned...and if my goal was easy...like a 20 watt rms continuous into 8 ohms 10 Hz to 15 KHz...and distortion not to exceed 1 %....I could handle that and love to play... Maybe I will look up good old classic audio amp and see their circuit...like a Carver or McIntosh...probably spelled wrong... Some how I am intrigued...the Tek design +5 with their Hfe custom picked gain...that it didn't work....yet not burned out...and you bailed...and I don't blame you one bit... We love our 7805, 7812...7815...LM350 linear power hogs, quiet, simple and stable.... Thanks again... After I get the axe or retire...I still like to Patrons or see if you could help...sort of bam...quick consult work...and that ball is in my court Thanks for reading.. Time to see and read our other more gifted electronic wiz follower comments... Kind regards Fred
Sziklai: Very good pronounciation. The sz is a digraph and results in a sharp s sound, the kind of sound English speakers naturally tend to make when they encounter an sz digraph. the a and i are pronounced seperately, so you did well. The a is acceptable but if you want to improve: it is almost the same sound as an english "uh" makes. just make sure to pronounce the i at the end. don't swallow it. it is almost as if there is a "aji" at the end, just without the "j". Btw, pretty much every name that starts with "sz" is either Hungarian or less likely: Polish. Annoyingly enough the "sz" in Polish sounds like a "sh", pretty much all digraphs in the languages are reversed.
Claim? Linearity is different, properties are different. The comparison is biased, the first transistor should have a resistor connected in the emitter in Sziklai.
Hungarian ( and Finnish) are not similar to other European languages; some people speculated the original peoples that founded the two countries came from far to the east. A looong time ago.
Finnish and Estonian are more closely related, with Hungarian a more distant (and not mutually intelligible) relative. They form the group of the Fino-Ugric languages, probably of Siberian origin.
I have most often seen Sziklai pairs in the last stage in audio power amps, with a small-signal PNP as the input of the pair and a power NPN at the output. The combination behaves like a PNP but has the current rating of the NPN. This can be a cheaper way to get a power PNP transistor, or in the old days maybe the only way.
Congrats on the patents! That's an impressive body of work.
Yep in the old days of the 1970s this is how we'd get class B or AB final stages going, using the classic 2N3055. A PNP complementary to it was not available (or rather very hard to get, the PN2955) so the stage was quasi-complementary: a Darlington with another medium-power NPN+2N3055, and a Sziklai with a medium-power PNP (much easier to find) and a 2N3055.
The good old Quad 405 used Sziklai pairs - I've renovated a fair few of them, still a good sounding amp, considering!
i was about to type that
still if you making negative ground devixe this is een better, could be a substiture for germanium darlington with no bias change
FYI:
Just an 'S' is like a 'SH'
Just a 'Z' is a soft 'S' (or like a 'Z')
'SZ' is a sharp 'S'
'CZ' is like a 'TS'
and 'CS' is like a 'TSH'
Except that the inventor of this circuit, George Clifford Sziklai has a Hungarian last name and in this case 'SZ' should be pronounced like 'S' in the word 'see'.
@@SLeslie : That's exactly what I wrote: 'SZ' is pronounced like a sharp 'S'.
Not sure where you're getting at... My FYI is indeed about the way Hungarians use and combine 'S', 'Z' and 'C' and what the results would be pronounced like in English.
@@bikkiikun Sorry, I probably misunderstood what "sharp S" should mean then. Although I believe I do know how Hungarians combine letters being a Hungarian myself
@@SLeslie : Understood.
Sharp 'S' is indeed, like you said, like in the English word "see".
A Soft 'S' would be like "Zombie" (in English).
@@bikkiikun Am I mistaken if I believe you to be German? Does the "soft S" come from the German like pronunciation like Simson (the former moped manufacturer in Suhl) for you? Fun fact: Those mopeds were popular here and we definitely pronounce its name improperly the Hungarian way with sharp S.
What I probably should have suggested from the beginning is if someone is debating whether he/she pronouncing a word right is to copy-paste it into google translate and read the word out loud with it.
The reason why the Vbe for the Sziklai pair is only 544mV is that the base current was 6.1mA/11,100 = 550nA. That's not an unusual voltage drop across a silicon p-n junction at such a tiny current. The Vbe's that you were expecting, 600mV - 700mV, are more typical of p-n junctions passing milliamps.
FYI a close approximation to Sziklai in general american is SEEK-la-yee, with the weight on the first syllable (but still short). The 'y' is a pretty good approximation of moving from the short 'a' to the short 'i'. It basically means "Of [a] boulder", which is what I wanted to bash my head into when I first started learning about transistors after basic circuit theory.
Or "seek lie", at least with an american accent. "Lie" is pronounced like "lai".
@@rich1051414 Hungarian doesnt merge vowels into full dipthongs, they're used as a transition when needed. 'Lie' omits the trailing 'i' which is a case suffix (the language is agglunative, so all cases are basically either prefixes or suffixes instead of seperate words). 'Lie'+'ey' might be a close approximation but its still a little off to my ears, mostly because the short 'i' is a very regional thing in English use IMO.
But honestly it's splitting hairs, any attempt is a welcome attempt considering how little people are even exposed to the language's existence.
seklae. greetings from Hungary. ps: you can listen it with google translate.
Big difference in the "on" voltage of the two configurations. The darlington cannot go lower than the EB voltage of the transistor while the skezke configuration can saturate to millivolts.
Agreed. And in many ways, the most important difference in switching applications.
on the flip side, and possibly more importantly, the Sziklai cannot sink below the pnp stage's Vbe
@@emiliaolfelt6370 AH! I had overlooked that! Thus, both circuits are lossy in a saturated state.
@@emiliaolfelt6370I consider Darlington better as emitter follower and Sziklai as common emitter.
You have 60 patents, awesome! What are some of your best?😮
That's why he's retired ;) but most of his patents must have been filed at HP which owns the parents and gives out a little fixed payment when filed or granted. Also, some patents would be associated with more than 10 people.
@@bayareapianist The corporations will give maybe 10$ per year per patent to the hard working engineer right?
in the US you get zero dollars. if you are lucky they might give you $100 or $1000 when they file it. in other countries I would be able to retire sooner, you get a good percentage. some US companies give percentages too, just not the ones I worked for.
@@IMSAIGuy Seriously impressive sir! 60 patents is incredible! 👏🏻 Additionally, your videos are very educational and I thank you for that. Keep up the great work.
Too bad you didn’t see any pay out after so many patents. I don’t see any either. Not even a gift card lol
It would be nice to see the curves of both types of transistors in curve tracer.
Ditto!
they look just about the same as an npn, save for the higher on voltage for the darlington, and a very slightly lower on voltage for the Sziklai.
Yep, would love to see Vce ensemble for Sziklai.
@@emiliaolfelt6370
Doug Self plotted them and the Skizklia pair has a much sharper turn on and gives lower distortion
@3:50 It's 0.544V instead of 0.6V BE voltage due to the very low base current. The diode characteristic is a e-function. If the current is lower than usual you move more to the left in the e-curve.
That was great! I had never seen a Sziklai pair. Thanks!
NPN silicon going to a Germanium PNP transistor Sziklai is a fun Sziklai for Germanium Fuzz guitar pedals. It greatly reduces the leak from the Germanium transistor in the pair.
The 'leak' would get more severe the harder the base NPN is pushed. Would that result in a dc bias when driven very hard? I guess it wouldn't matter if pass through a capacitor to remove any bias anyway.
@@rich1051414 yep, you got it.
In my experience, if the pair isn't too thermally sensitive and you can find a reasonable operating point, you can make a fuzz out of it that someone will find interesting! To that point- in college, in 1998 or so, I built some fuzz faces for friends with used germanium from a transistor radio, by tweaking the resistor values until they sounded great. One in particular sounded amazing. I didn't have RG keen's measurement rig yet, so my tester showed the input transistor had a gain over 200. I'm certain it was just leaky. Whatever the case, I managed to bias it up by ear and the results were great. I guess, in the end, there are no rules as long as it sounds good!
2:30, if the hFE of a transistor is 173, let's round it to 200, so putting 2 transistors in darlington pair configuration should give 200 * 200 = 40,000. Or roughly 20,000 to 60,000 not 862,000. What's going on?
I may be wrong, but I think a Sziklai would be more temperature stable than a Darlington. PNP and NPN transistor combinations are used to stabilize temperature drift of control voltage in synthesizers fairly often because they behave oppositely to an increase in temperature.
The Sziklai is usually implemented with a bias resistor on the second transistor and there are 2 topologies. One disadvantage of the Darlington is its low frequency response, compared to a common emitter stage. While an emitter follower stage, using a 2N3904 for example, can have a frequency response of 2 MHz, a Darlington stage will only operate into a few tens of kiloHertz. The Sziglai has a better frequency response than the Darlington,. I haven't done a calculation (sum of the inverse time constants) for the Sziglai so can't give a more precise answer. The cascode and differential amplifiers, which are both 2 transistor stages, have much higher frequency responses (~12 MHz) and higher gain than a common emitter amplifier.
I came up with this configuration on my own when trying to avoid the inability to saturate a Darlington pair unless you have a voltage at the first base 1.2V above the second transistor's emitter. It just seemed like a very simple complement to a Darlington, and handy when you wanted to fully enhance a PNP power transistor, but didn't want to expose the driving circuit to the potential high voltage on the PNP's base (ie logic).
Also the Sziklai pair is the basis of your IGBT power device, integrating a high voltage low gate capacitance n channel mosfet with an absolute brute of a PNP power transistor, and a diffused base emitter resistor, plus a massive body diode thrown in almost for free as well. Low drive needed for the mosfet, and it can turn on and off fast, and the power PNP transistor can suffer the miller capacitance without reflecting into the gate as a huge charge you need to either insert or remove. So the advantages of a mosfet of not needing huge charge to drive for a huge current, but also no huge gate capacitance either, even if the overall device is relatively slow in acting, but it does have very good pulse rating and also a very low forward voltage drop, unlike the darlington, which will always have around 1V5 of drop across it irrespective of gate current, due to the 2 base emitter junctions, and the first transistor needing to have some voltage across it, plus the second transistor will never saturate.
In 1958 IBM actually employed more engineers than the "Lazy L" as it was called at the time. IBM was developing the secret Winchester Drive at that time, replacing the drum drive.
Great discussion, glad to have this shown to the masses! Maybe some enterprising young engineers will consider the Szilklai in their own designs. I see this in a lot of AF/RF amps due to its ability to work at lower bias levels. I'm not sure how much of that is due to actual circuit knowledge or just copying libraried designs.
An interesting topic that might be fun to cover is the difference (probably no more) between Japanese 'electron-flow' teaching and western 'voltage flow' - *and* how that influenced the design of amplifiers, increased the use of PNP transistors in Japan, and how it makes 1960s-1980s Japanese schematics so darn hard to figure out ;)
Great video, love the content. Cheers!
03:52 that's because Vbe determines the Ic of a transistor, and in this case the collector current for the NPN tranistor would be reaaaally small (since it's the Ib for the PNP that has a current of 6mA trough it). Even in a darlington the Vbe of the driving transistor will be smaller than that of the power transistor.
The Sziklai design is very convenient for amplifier power output stages. At first you don't lose less output voltage swing, because it can be driven closer to the + and - power rails. At second the temperature compensation of the idle current of the output stage is less critical because the driver transistor gets less hot if you oversize it a little bit (use a 500 mA to 1 A device if you have to handle 50 to 100 mA) and don't locate it in the direct vicinity of the output transistors.
It is also good practice to use feedback resistors in the Darlington or Sziklai designs to limit the total current amplification to a reasonable amount (1000 to 2000) in order to get a more linear characteristic of the emitter follower output stage.
Built one of those IMSAI 8080 computers in 1976. Built 8 4KB memory cards for a whopping 32KB. We bought a Z80 board when those became available and on a good day you could flip the switch to 4MHz!
So if your aim is for power, I imagine the two transistors don't have to be identical or complementary, I think perhaps the "gate driver" transistor can be a signal transistor that's driving a power transistor.
Since in both configurations the resulting Beta is approximately the product of both transistor Beta values it can be useful to pick a large beta one for the smaller one, as it needn't be rated for the output current, nor heat dissipation ability
Sziklai is also useful for using NPN power transistor and PNP signal transistor as a PNP power transistor.
At 1:09 you have pronounced his name pretty mutch correctly, it's ' Sikla'y ' . The 'sz' in Hungary is meant to create the 's' sound while the vocal letters do not merge in 1 sound.
The times I've seen Sziklai pairs being used is in power amplifiers before silicon transistors. It was significantly easier to make germanium PNP power transistors than NPN ones and with a Sziklai pair you could omit the use of hard to get and expensive power NPNs :-).
I also believe that the darlington is slower. There is a capacitance Cbe that can charge quickly if the base voltage rises but then you don’t have a good way to discharge it on the way down.
Interesting. I've been wreaking electronics since the 1970s, and this is actually the first time I've ever heard of Sziklai pairs. They're intriguing, but not overly handy.
Here is my observation...
- Sziklai is pronounced differently in USA and the rest of the word. When I moved to USA I had to relearn to pronounce a lot of words such as Boston which is pronounced "Baas ten" here.
- Here is shown an NPN driving a PNP (power) transistor. Power silicon PNP were harder to make and were much more expensive. For example MJ2955 was many time more expensive than its NPN counter part 2N3055. The Szklai pair is counter to reality in silicon age. But one thing we forget is Sziklai came up with this combination in germanium eara. When he invented this circuit in 50s, it was after transistors had been invented and they were mostly made with germanium. In that eara, PNP germanium power transistor were more common than NPN. I have never seen an NPN power transistor myself, and I have seen a very few low power NPN only. That's why you always see an NPN driving a PNP transistor when someone describes the pair.
- Both of these pairs have their issues like the frequency responses/the unity gains frequencies, and saturation of driving transistor. That's why you only see them at the power output stages of audio amplifiers. Another more common circuit is using both in common emitter and driving the base of power transistor with a resistor. That resisitor limits the max current to the power transistor.
- I'm trying to earn 3x1 cents here ;)
It's very interesting. In Hungary the used name of "Sziklai" is composite switched transistors or simply composite transistor. We never use the name "Sziklai".
All your videos are awesome :)
The hfe given by TC1 for darlington seems to be quite high : 800k or 600k ! That means that each transistor has an hfe of about 800 to 900!
To be fair, I'm not all that sure a simplistic "multitester" like that, with next to no analog front end at all to work with can be expected to resolve a hfe like that beyond "oh shit it's really high".
Thanks for the video. The video reminded me of the resistance measurement current source used in the HP34401A DVM. For the life of me I can't figure out how this circuit works. This circuit would make a great video episode. Part of the circuit contains two strings of 4 PNP transistors connected in series, an op amp and four selected resistors to set the current.
Craig
To help a little - a number of European languages have letter combinations that seem horrendous in English! e.g.Polish, SZ is usually pronounced as SH (Shh, Quiet) in English. CZ is usually pronounced as CH (Chalk) The letter combination SZCZ (Oh wow!) is not too bad if you think of an English word 'Ashchurch' A suburb, - not too bad, eh? So, Can I suggest Sziklai could be pronounced as Shik-lai. Hope this helps. Best regards. Excellent lecture, by the way.
Thank you.
I didn't know about the Sziklai variant before, interesting 😮
I've heard that Sziklai on modern HF transistors can start to oscillate because one transistor is in CE.
Well presented 👏👏
An interesting topic at some point in time would be some of your patents. Maybe make up a list of them so we can pull them up on Google Patents. Very interesting indeed.
I created an abstract from the transcript using Gemini Advanced 1.0:
*Abstract*
This video explores the configurations and applications of two transistor types: the Darlington and the Sziklai (pronounced 'Sik-lie'). The speaker begins by emphasizing the similarities between the two, noting they both serve to amplify current. The primary difference lies in their base-to-emitter voltage (Vbe). A Darlington configuration will have approximately double the Vbe of a single transistor due to the cascading of two transistors. Due to their lower Vbe, Sziklai transistors can be advantageous when used with low-voltage microcontrollers. For higher current applications, the speaker notes that NPN transistors remain more readily available and affordable, explaining the continued popularity of Darlington pairs.
*Keywords:* Darlington Transistor, Sziklai Transistor, Vbe, Current Amplification, Microcontrollers
0:25 szikla=rock +i mark of origin. like newyork-i mean somebody form newyork. used as surname at medieval ages as house name to show where is the "headquarters" of the given noble family. but the szikla is not a city... so its probably a hungarised name form late 1800's (probably shaul) and not follow the naming conventions.
@2:13 Funny how you still managed to screw up the value even with the correction. 😆
Interesting video nonetheless, had not heard of a Sziklai Transistor before.
I am studying in Hungary, it simply pronounced Siklai
There also should be difference in input/output capacitance, i guess.
Sziklai pairs seem to love to oscillate. They're not as useful as you'd expect.
Something like Seek-la-ee
Another Hungarian checking in -- I was about to comment the same thing.
Well that's a first for me. I cannot believe ive never encountered this before.
Interesting topic. The first transistor has a much lower base current and as such also a low Vbe which is a nice feature.
I’m thinking what would be the behavior over frequency as the miller caps may work differently in the Szillai pair. I need to check this in LTSpice. Any thoughts anyone?
I would think the Sziklai would be able to de-saturate faster as a small resistor could be added in the middle to pull the smaller transistor out of saturation.
thank you so much!
I will appreciate if you explain the functioning of a TL431 connected to the ground of a 7805 regulator to get a variable supply from 1 to 5 volts .The TL431 refrence pin is conected to a voltage divider between the 7805 output and ground .Thank you very much
that won't do. In such a circuit the voltage at the 7805 ground pin will be above ground. the 7805 output voltage wil be 5 volts above it's ground pin.
However the 431 will sink current at the cathode until it's ref input drops below 2.5V. To regulate to 2.5V at ref input the 7805 output must be at 2.5 volts times the resistor divider ratio, which is achieved when the 7805 ground pin is at 5 volts below that.
Within the 7805 some current flows from the input pin to the ground pin, which pulls the cathode of the 741 until the output voltage raises to the level that results in the 2.5V at the reference pin of the 741
maybe this would help: th-cam.com/video/oYj67rNUUNk/w-d-xo.htmlsi=hYq-Qia5rkE1_KKH
I know the Peak Atlas will identify Darlington transistors... and I believe that some of the other low cost (probably Chinese) component testers identify Darlintons too. (... although it is probably just a flag in the firmware, when the current gain goes beyond a certain value... say 300?)
Another way they are different is in de-saturation time.
You forgot to mention that the Sziklai (my guess tsiklaiy pronunciation) configuration also has a much lower output saturation voltage, as it's about (1 x Vbe) + (1 x Vsat) where as for the Darlington it is roughly (2 x Vbe) + (1 x Vsat).
Vbe depends on the current so it is often less than 0.6 V
Why did the hFE for the Darlington change the second time you pushed the button?
those little boxes don't have much resolution at those very high beta numbers
Probably a one bit change, IIRC its a 8 bit ADC in these little testers.
I learned the Sziklai circuit as the "White Darlington". I have no idea who What was. If anyone does, please let me know.
BTW: Mr Sziklai's circuit can be done 3 transistors deep. A PNP driving an NPN driving a PNP can give you a really high HFE.
@@seanoconnor1984With high value emitter to base resistors, they also work to control heavy loads from low power circuits. I did the 3 transistor thing in a power supply I made for hobby stuff. On PNP drove 2 NPNs which drove 4 PNPs. It spread the heat out enough to let me get all the heat into a big heat sink.
Sziklai. Say it with just an S in english.
Never knew about this guy.
I studied electronics for years ( high school ) but i never heard about Sziklai ( Sz is just S for english ) ... or i just dont remember this guy from the books !
I just tried signing on to you Patron to join your account there, but patron poeple are making it difficult.
Interested in the tool you used to ID the chips. I am sure I can find it elsewhere (Walmart, Amazon...), but I don't want to give my money tho those things, I wanted to purchase it from you. ...
I don't sell them. Patron is just a way to send me money to support the channel
Thanks 👍
That`s the easiest part of Hungarian:) SZ it is one letter in alphabet two character. It stands for S in English. SIKLAI
(S in Hungarian stands for SH in EngliSH)
Hey there!
Great video! I haven't heard of the Sziklai pair before.
If your trying to pronounce Sziklai better the 'Sz' written in hungarian makes the regular 's' english sound, the 'i' is like in 'hit', the 'k' is nothing special just like in 'kid', again the 'l' is just a regular l like in 'literature', the 'a' is pronounced like in 'around', and the last 'i' is pronounced just like the first one. Or, y'know, you could just use the google translator instead of reading my blabbering...
Btw if someone is still reading this 'Szikla' in hungarian means big rock or boulder and it is also a name of a village. If you add an 'i' to the end of the word you'll get 'Sziklai' which means 'from Szikla' just like the name of Ludwig van Beethoven or Leonardo da Vinci.
Didn't know about this topology, I think this topology is being used in the voltage regulator circuit of my NAD 3140 amplifier... but I couldn't really make sense (not an electronics guru) of this part of the circuit, so might make more sense now... not sure why they didn't just use NPN/PNP darlington pairs as the circuit is dropping 10+ volts so 1.2 vs 0.6 vbe wouldn't have mattered, not sure why a voltage regulator would need the high gain of these topologies either 🤔
Both these configurations are great, just beware if they are used somewhere where they can "see" RF signals, they are a pain in the behind to stabilise. Don't ask how I know 😑
@@seanoconnor1984 Exactly! And emitter followers can be biased as to seem inductively reacting. I used that quite often to taylor freq. responses of circuits, but any inaccuracy in the extraction of the parasitic caps of the layout means a few hours in the lab debugging the circuit ... As a rule of thumb: an amplifier oscillates, and an oscillator amplifies, thanks Murphy :-)
Nobody can pronounce Hungarian names. It's impossible. 🙂
While I have long known about Darlington transistors, this is the first time I've heard of the Sziklai.
BTW, I used to have an IMSAI 8080. I bought the kit in Nov. 1976. I remember people asking what would I use a computer for. 🙂
THE ANSWER TO 1984.
Thanks alot....
I knew you are a genius, a great inventor and HP was a good place where they encouraged IP...
You have 57 more than I. At least we got like $2K each & better than a kick. We had a few weirdos...then liked to patent strange things...just for the money...but had little practical purposes.....I mean I admired their creativity...
I am not even in that league of brilliant inventors...
There was popular little Darlington TO-92 we buy by in the 1000's...plastic
I have to look it up..
I think called an A10...Motorola
Eariler during your Tek scope...and you gave up on the semi discrete design and replaced with 7805...
I know I babbled...later it looked like there was another voltage...for I thought just 11 VDC dropped to 5....then I heard 19 VDC...
In this Tek little circuit...
It seemed to me..the comparator..if like a LM339..wouldn't it just switch on or off...and be noisy and output chopped then need filtered...so the comparator in the Tek... isn't it really a linear bias little Op amp....?..not a noisy switcher
I like a better recording instead of my tiny phone...at work big display....no time to look up Tek scope +5 VDC or watch your TH-cam lession ...
Probably head to Verizon...for a tablet...
Our like fast curve tracer or computer that basically stored every bias point in digital...and list pass fail ....we had some Tek carts too just like yours on carts for nice entire curve display....Our main transistor tester was a Lorlin driven by the PDP-11...about the size of a small closet....plus the transistor handlers....
I mistakenly I called it a Loran...which ia a navigation system....
Our little AA community college....for those circuits 1986 ish we had the Malvino book...no calculus needed -I could hang with that...sort of simple and appropriate for the class
I seen the Sziklai pair with the Darlington.....in audio amps...National Semiconductor great little old school AB bias...nice application notes....would show like what Malvino touted...
NPN Darlington for top pair...and Sziklai pair for like the bottom or when the signal goes negative....the advantage being the Darlington and Sziklai when paried they both used both NPN in final pass with 3 diode....so crossover maybe less
Malvino annoyed me somewhat...no real circuits from the standpoint of use a 2N5179...I like cracking it open...time to time...
And NPN Darlington on top with PNP Darlington on bottom...is usual...but maybe tougher to find NPN PNP complementary pairs...plus there were 4 diodes...to help bias...
So cross over distortion must be worse when your final audio amp uses NPN Darlington on top and PNP Darlington
pair on bottom.....4 compensation diodes
I thought he was Japanese too!
Thanks for the enlightenment of his accomplishments and reminding us of yours...
Another reader asked a few of say your biggest most successful..but then you might be worried we might look up your name...
When I worked at IBM...to install and test semiconductor tools...they had great patent reward system...
Luckily...I never tasked to build...soild state audio amp...which I like to do if assigned...and if my goal was easy...like a 20 watt rms continuous into 8 ohms 10 Hz to 15 KHz...and distortion not to exceed 1 %....I could handle that and love to play...
Maybe I will look up good old classic audio amp and see their circuit...like a Carver or McIntosh...probably spelled wrong...
Some how I am intrigued...the Tek design +5 with their Hfe custom picked gain...that it didn't work....yet not burned out...and you bailed...and I don't blame you one bit...
We love our 7805, 7812...7815...LM350 linear power hogs, quiet, simple and stable....
Thanks again...
After I get the axe or retire...I still like to Patrons or see if you could help...sort of bam...quick consult work...and that ball is in my court
Thanks for reading..
Time to see and read our other more gifted electronic wiz follower comments...
Kind regards
Fred
You could also use the opposite of that, PNP and NPN. Don’t know what it’d be called though!
It's still called Sziklai regardless if it's PNP+NPN or NPN+PNP. Just like it's still a Darlington if it's 2 PNPs instead of 2 NPNs.
why is the hFE so high? (800k) I expect it to be hFE * hFE, which is 200 * 200 = 40k
no load and low current testing
Sziklai: Very good pronounciation. The sz is a digraph and results in a sharp s sound, the kind of sound English speakers naturally tend to make when they encounter an sz digraph. the a and i are pronounced seperately, so you did well. The a is acceptable but if you want to improve: it is almost the same sound as an english "uh" makes. just make sure to pronounce the i at the end. don't swallow it. it is almost as if there is a "aji" at the end, just without the "j".
Btw, pretty much every name that starts with "sz" is either Hungarian or less likely: Polish. Annoyingly enough the "sz" in Polish sounds like a "sh", pretty much all digraphs in the languages are reversed.
It's something like [sheclay].
Claim? Linearity is different, properties are different.
The comparison is biased, the first transistor should have a resistor connected in the emitter in Sziklai.
Cool 😀
Great!
Hungarian ( and Finnish) are not similar to other European languages; some people speculated the original peoples that founded the two countries came from far to the east.
A looong time ago.
Finnish and Estonian are more closely related, with Hungarian a more distant (and not mutually intelligible) relative. They form the group of the Fino-Ugric languages, probably of Siberian origin.
The ( Sziklai ) used to be called [ super α pair ] or , super alpha pair ... easier to pronounce ? .... DAVE™🛑
seeklahee
Hungarian is goofy. sz = s, s = sh, so its just pronounced sik - lai