# EHT power supply design and construction

#### Aleph(0)

Joined Mar 14, 2015
597
HP ok so here's problem germane to thread that's totally been driving me crazy! So below is diagram of LV PSU ripple filter.

So for given values, resonant Freq of L1-C2 ckt is abt 37.5 Hz which is no problem at all! But I definitely see potential for huge problem on two _fronts_.

1) If L1 is overloaded into saturation it could lower L1's inductance enough to bring resonant freq near 120Hz which would cause HUGE (like multi-kV) drop across C2!

2) Inductive load could tune out enough of C2's capacitance to cause same problem!

So FWIW L2 would need to be lowered to ≈ 117.27μH to resonate with 15mF C2! Also it would only take ≈ 130 MICRO-Henries of load inductance to tune C2 down to ≈ 1.466mF bringing it to 120Hz resonance with 1.2mH L1! Also plz remember just getting anywhere near resonance is bad enough!

As for me, I am perpetually incapacitated by Aleph(0)'s unspeakable pulchritude. *sigh*
@KL7AJ when you say things like that it's like ur saying my thoughts and ideas are totally unimportant! I don't like being objectified and I bet you don't either! So I know ur not being deliberately rude and I prolly brought it all on myself with _numpty act_ but if it ever was funny it's totally run its course! So I'm asking you to plz cool it with silly remarks and contribute something of substance cuz we definitely need all the help we can get with all we're trying to do here!

#### Hypatia's Protege

Joined Mar 1, 2015
3,226
HP ok so here's problem germane to thread that's totally been driving me crazy! So below is diagram of LV PSU ripple filter.

So for given values, resonant Freq of L1-C2 ckt is abt 37.5 Hz which is no problem at all! But I definitely see potential for huge problem on two _fronts_.

1) If L1 is overloaded into saturation it could lower L1's inductance enough to bring resonant freq near 120Hz which would cause HUGE (like multi-kV) drop across C2!

2) Inductive load could tune out enough of C2's capacitance to cause same problem!

So FWIW L2 would need to be lowered to ≈ 117.27μH to resonate with 15mF C2! Also it would only take ≈ 130 MICRO-Henries of load inductance to tune C2 down to ≈ 1.466mF bringing it to 120Hz resonance with 1.2mH L1! Also plz remember just getting anywhere near resonance is bad enough!
First of all -- thanks for the on-topic inquiry!

I will offer a point-by-point reply when time permits --- For the nonce, as a relevant exercise in perspective, I ask you to consider the difficulties attending comprehension of your right hand's viability sans consideration of said member's relationship to your body as a whole -- get it?

Best regards
HP

#### Aleph(0)

Joined Mar 14, 2015
597
HP I'd totally be throwing stones from glass house making smart a$$remarks about _outside demands on your time_ So instead of that, here's things for you to help with WHEN U DO HAVE TIME! 1) Ok so I 'did math' like you said which totally proved max efficiency of 1Φ bridge rectifier is like u said 8/Pi^2 so like 80% but that's POWER not current! And anyhow (for benefit of other readers) here's setup: 120VRMS line→Variac→2:1 Stepdown xfmr→Bridge rectifier→Pi section ripple filter (15000μF-1.2mH-15000μF) →Load (3.4Ω 5kw resistor which has inductance of 34μH [so XL of 25.6 mΩ at 120 Hz which I say is totally negligible])! 2) Anyhow real world test results for that setup are I_sec/I_pri ≈ 1.35 which I say is disappointingly weird but E_sec/E_pri ≈ 0.543V (which is totally expected for Pi section under hvy load)! But current is pathetic! Now it's not inefficiency in xfmr cuz just xfmr w/ unloaded secondary has primary magnetizing current of only abt 20mA! Also unloaded bridge basically draws no current at all once Caps are charged! Also here's another irritating thing! Pri AC current varies 0.6A to 1A based on which 180° position clip-around ammeter is over wire! So being clear I'm talking strictly AC meter (so really just current transformer) and problem happens with dirt cheap HF all the way up to Fluke meters (on AC setting)! So here's what I've checked! 1) Waveform is PERFECT sine wave! 2) There is absolutely NO DC offset on line (I checked even though it comes though dedicated 10 KW Iso xfmr) 3) There is no noise on line w/ greater magnitude than 40pV! 4) Meter clip is 90° to wire I ask you to consider the difficulties attending comprehension of your right hand's viability sans consideration of said member's relationship to your body as a whole -- get it? So HP anyhow I get your point abt illogic of looking on pi-section as cap plus L-section rather than _big picture_ (which I definitely want to discuss) but for now these other problems are bugging me even more Also abt Theo! So since she confessed (I say she was caught) in _ flagrante delicto_ with Dept head's husband, she's to busy just maneuvering to keep her position (npi) to post on here for now! But since Antenna & transmission line theory are subject of general interest to almost all electronics pros and hobbyists I say u could answer her _2'nd hand' questions_? So here's very sincere disclaimer to mods! I'm not _telling tales out of school_ abt Theo cuz she's totally _confessed_ to affair and anyhow she's one of my best friends so I'd never say or do anything to hurt her! Last edited: #### Hypatia's Protege Joined Mar 1, 2015 3,226 HP I'd totally be throwing stones from glass house making smart a$$ remarks about _outside demands on your time_
Methinks you may be on to something there...

2) Anyhow real world test results for that setup are I_sec/I_pri ≈ 1.35 which I say is disappointingly weird but E_sec/E_pri ≈ 0.543V (which is totally expected for Pi section under hvy load)! But current is pathetic! Now it's not inefficiency in xfmr cuz just xfmr w/ unloaded secondary has primary magnetizing current of only abt 20mA! Also unloaded bridge basically draws no current at all once Caps are charged!

First off -- let's not confuse rectifier/filter output current with transformer secondary current! -- Some readers are 'fussy' about that sort of thing (pointing directly at self)...

But current is pathetic!
And again - your haste-induced 'blithering' requires clarification!

What you mean is that --for a given rectifier/ripple filter DC output current-- the AC current apparent upon the step-down transformer's primary is higher than you'd like (CIP ≈ 0.74*IRectOut ) --- No! I'm not 'splitting hairs'! --- Stated your way - one might --quite despite your reassurances to the contrary-- be forgiven for indicting transformer efficiency! In point of fact 'shorting' the secondary results in currents ranging to several kA -- Pending opening of the load-center OCP interrupter...

But to respond to your intended question:

It may behoove you to bear in mind that real-world electrical 'filtration' entails (in addition to 'recycling') some energy loss - even in LC filters --- Again, do the math! - Integration is painless! FWIW I'm guessing a 'perfect world' Iprimary:IRectOut figure = √2/2 but then I didn't do the math

Also here's another irritating thing! Pri AC current varies 0.6A to 1A based on which 180° position clip-around ammeter is over wire! So being clear I'm talking strictly AC meter (so really just current transformer) and problem happens with dirt cheap HF all the way up to Fluke meters (on AC setting)! So here's what I've checked!

1) Waveform is PERFECT sine wave!
2) There is absolutely NO DC offset on line (I checked even though it comes though dedicated 10 KW Iso xfmr)
3) There is no noise on line w/ greater magnitude than 40pV!
4) Meter clip is 90° to wire
Such has been my observation as well -- FWIW Said phenomena is most pronounced when 'looking' at the primary current of a restively loaded transformer... Nor is a significant difference in said discrepancy seen to exist between 'True RMS' vs 'standard' indicators (which being as one would expect considering the near-pure sinusoidal current characteristic)...
It's... It's..... vaguely reminiscent of a ditty crudely paraphraseable as....

It's everything I expect
It's everything I deem
It's everything a lifetime's experience tells me it should be
It's exactly the right waveform
At exactly the right parameters
But it's nowhere near consistent...
And I don't know why!

<<<With profound apologies to 'Vertical Horizon'>>>

Seriously! Should anyone have insight into this - I implore you to share!

Also abt Theo! So since she confessed (I say she was caught) in _ flagrante delicto_ with Dept head's husband, she's to busy just maneuvering to keep her position (npi) to post on here for now! But since Antenna & transmission line theory are subject of general interest to almost all electronics pros and hobbyists I say u could answer her _2'nd hand' questions_?

So here's very sincere disclaimer to mods! I'm not _telling tales tails out of school_ abt Theo cuz she's totally _confessed_ to affair and anyhow she's one of my best friends so I'd never say or do anything to hurt her!
While I wholly concur with your assertion that @theodoravain will not likely take issue with the above quoted text -- I have to wonder how turning this thread into a tabloid furthers your oft-stated goal of mainstreaming same?

Best regards
HP

#### KL7AJ

Joined Nov 4, 2008
2,225
HP ok so here's problem germane to thread that's totally been driving me crazy! So below is diagram of LV PSU ripple filter.

So for given values, resonant Freq of L1-C2 ckt is abt 37.5 Hz which is no problem at all! But I definitely see potential for huge problem on two _fronts_.

1) If L1 is overloaded into saturation it could lower L1's inductance enough to bring resonant freq near 120Hz which would cause HUGE (like multi-kV) drop across C2!

2) Inductive load could tune out enough of C2's capacitance to cause same problem!

So FWIW L2 would need to be lowered to ≈ 117.27μH to resonate with 15mF C2! Also it would only take ≈ 130 MICRO-Henries of load inductance to tune C2 down to ≈ 1.466mF bringing it to 120Hz resonance with 1.2mH L1! Also plz remember just getting anywhere near resonance is bad enough!

@KL7AJ when you say things like that it's like ur saying my thoughts and ideas are totally unimportant! I don't like being objectified and I bet you don't either! So I know ur not being deliberately rude and I prolly brought it all on myself with _numpty act_ but if it ever was funny it's totally run its course! So I'm asking you to plz cool it with silly remarks and contribute something of substance cuz we definitely need all the help we can get with all we're trying to do here!

View attachment 181649
Have you modeled L1's saturation characteristics? LT spice assumes an ideal inductor unless stated otherwise

#### Hypatia's Protege

Joined Mar 1, 2015
3,226
Have you modeled L1's saturation characteristics? LT spice assumes an ideal inductor unless stated otherwise
AFAIK @Aleph(0)'s calculations are based upon parameters derived from her profiling of the 'real-world' components... In any event I'll alert her to your reply

Many thanks!
HP

#### Aleph(0)

Joined Mar 14, 2015
597
@KL7AJ Huge tnx for posting serious content!

Have you modeled L1's saturation characteristics? LT spice assumes an ideal inductor unless stated otherwise
Kl7AJ it's like HP says I'm just _crunching math_ thinking on possible problems with real parts. (Being clear I used LTS schematic editor just as drafting tool w/o running simulations). Anyhow if I get time I'll just _breadboard_ it instead of trust emulator (which sometimes _lies like rug_ for reasons NOBODY can explain) so for simple ckt like ripple filter I say it's best to just breadboard! That way if it doesn't work as expected I totally know problem is MY thinking cuz reality is the only perfect _emulator_

So anyhow two main problems for me is I'm totally like _RF oriented_ so operation of ckts at vry low freqs (like 120Hz) and DC aren't like _second nature_ to me at all Also as example, thinking on effect of DC bias on ferromagnetic inductor is basically _new territory_ for me cuz it basically forces choke to be _magnetic amplifier/attenuator_ to AC part of pwr _signal_

First off -- let's not confuse rectifier/filter output current with transformer secondary current! -- Some readers are 'fussy' about that sort of thing (pointing directly at self)...
HP sry abt that! Where I was talking abt whole PSU (Xfmr pri input to RECTIFER+FILTER output) I was thinking right and typing wrong!

Such has been my observation as well -- FWIW Said phenomena is most pronounced when 'looking' at the primary current of a restively loaded transformer... Nor is a significant difference in said discrepancy seen to exist between 'True RMS' vs 'standard' indicators (which being as one would expect considering the near-pure sinusoidal current characteristic)...
HP since ur saying you don't understand discrepancy based on left vs right probe direction either, I say we should post question about it in _on topic forum_ cuz difference of like 5% to 20% (based on current level being measured) is way too much to just _sweep under rug_! So anyhow I'm asking you cuz since you can upload pix, readers can totally understand problem!

While I wholly concur with your assertion that @theodoravain will not likely take issue with the above quoted text -- I have to wonder how turning this thread into a tabloid furthers your oft-stated goal of mainstreaming same?
HP I know ur right but I couldn't resist after her nasty little _exposé_ abt me (on different site)

Last edited:

#### Aleph(0)

Joined Mar 14, 2015
597

JohnInTx said:
Closed by request.
Score: 2 pages, 7 deletions, 3 warnings and 1 ban.
It’s why we aren’t going to allow politics on the boards
Emphasis mine.

HP So you learned same thing as me with some of my poll threads! Which is just MENTIONING SUBJECT of politics w/o injecting politics itself still starts 'fire' that gets totally out of control no matter what!

So HUGE THANKS to @JohnInTX and all other staff who are part of the _WE_ in John's post I quoted above! Cuz AAC definitely doesn't need politics AT ALL!

#### Hypatia's Protege

Joined Mar 1, 2015
3,226
HP since ur saying you don't understand discrepancy based on left vs right probe direction either, I say we should post question about it in _on topic forum_ cuz difference of like 5% to 20% (based on current level being measured) is way too much to just _sweep under rug_! So anyhow I'm asking you cuz since you can upload pix, readers can totally understand problem!
Aleph! Really! I'm becoming a bit weary of this sort of 'thing'! -- My time is at a premium as well! -- You are quite capable of posting your inquiry to the 'Test And Measurement' forum sans 'visual aids'...

À la

"Why do AC 'clamp-around' current indicators tend to exhibit significant discrepancy each 180° 'clamp' position coaxial to the conductor carrying the current under test (granting a pure sinusoidal current waveform falling entirely within said indicator's specified frequency and amplitude ranges and sans DC offset)" --- See? Composition time Ca. a single minute and utterly painless! -- I'll wager you can do better!

HP I know ur right but I couldn't resist after her nasty little _exposé_ abt me (on different site)
If it is indeed your desire that this thread becomes 'on topic ready' -- Or, in any event, attains to the dignity of a resolute 'development area' for our tutorials and projects (which being composed and offered as serious educational resources), might I suggest you demonstrate same via course of conduct? -- Such does not include 'larking about' on the thread! -- Bad habits are best broken cleanly!

Best regards
HP

#### KL7AJ

Joined Nov 4, 2008
2,225
@KL7AJ Huge tnx for posting serious content!

Kl7AJ it's like HP says I'm just _crunching math_ thinking on possible problems with real parts. (Being clear I used LTS schematic editor just as drafting tool w/o running simulations). Anyhow if I get time I'll just _breadboard_ it instead of trust emulator
That's a good approach...Bob Pease, my hero said, "My favorite programming language is solder."

#### Aleph(0)

Joined Mar 14, 2015
597
So since HP had these _kind_ things to say abt me in a PM with another member I need to respond here to give my side to everyone else she's maybe also saying it to:
Frankly - between Aleph's absenteeism (owed, ostensibly, to VPN difficulties) and JC's photo-doxing of myself I'm bound to say the "fun's" gone right out of it (i.e. participation on AAC fora)!
(Emphasis Mine)

So anyhow here's my totally honest,verbatim reply!

_HP don't be laying that on my doorstep! All Zenmate servers are so totally and permanently DEAD they're prolly drawing flies! So like u know (but for benefit of others who don't) for now I'm using Avira's Phantom Vpn through your hub! But since Avira only allows 200Mb per day plz don't expect much! HP now I'm going to tell you something! You keep nagging on me abt being too paranoid to go _barefoot_ with my Xplornet IP for all the world to see! But I say that's total hypocrisy shaming me for privacy concerns when you had a major hissy just cuz of JC posting your piccy! So RU saying only YOUR privacy matters?

BTW <<Username redacted for privacy>> huge thanks for your concern! If everything works out we'll be back to our normal AAC schedule by late November!
_

HP BTW plz check your e for schematic of analog audio input preamps to FM broadcast multiplexer! I just don't get those criss-cross RC filters (output of each channel to input of other) cuz it seems like they'd increase crosstalk (and so reduce channel separation) for non-common mode signals? And just to say I don't mean preemphasis filters!

So I'm asking U to please upload schematic and explain and if you don't get it either plz upload schematic with question to appropriate On Topic forum! Sry 2b so dumb but op-amp circuits using asymmetrical supply (by which I mean VCC and GND instead of split supply) totally drive crazy just thinking on them

#### Hypatia's Protege

Joined Mar 1, 2015
3,226
@Aleph(0) --- I'll thank you to quote in context! -- Note that I freely owned my 'share' of the 'blame' as well -- to wit:
Then too our (Ahem, my) insistence upon 'disbanding' the 'beta test team' has left us in a bit of an 'echo chamber'
HP BTW plz check your e for schematic of analog audio input preamps to FM broadcast multiplexer! I just don't get those criss-cross RC filters (output of each channel to input of other) cuz it seems like they'd increase crosstalk (and so reduce channel separation) for non-common mode signals? And just to say I don't mean preemphasis filters!
It is indeed an interesting 'problem' -- at 'first blush' I have to wonder if the 'DC Path' provided by the parallel R/C arrangements are with the (additional) aim of dynamically biasing the inputs above ground (which being the 'negative' rail, as it were)...

op-amp circuits using asymmetrical supply (by which I mean VCC and GND instead of split supply) totally drive crazy just thinking on them
You and I both

So I'm asking U to please upload schematic and explain and if you don't get it either plz upload schematic with question to appropriate On Topic forum!
Perhaps this weekend, time permitting...

Very best regards
HP

#### Hypatia's Protege

Joined Mar 1, 2015
3,226
HP BTW plz check your e for schematic of analog audio input preamps to FM broadcast multiplexer! I just don't get those criss-cross RC filters (output of each channel to input of other) cuz it seems like they'd increase crosstalk (and so reduce channel separation) for non-common mode signals? And just to say I don't mean preemphasis filters!

So I'm asking U to please upload schematic and explain and if you don't get it either plz upload schematic with question to appropriate On Topic forum! Sry 2b so dumb but op-amp circuits using asymmetrical supply (by which I mean VCC and GND instead of split supply) totally drive crazy just thinking on them
Very well then -- Below is the audio-processing portion of the circuit (curious 'cross-channel' filters circled in blue).

As an aside, please note that I corrected a typo Re: the resistor value in the right channel's pre-emphasis filter... --- That said, North American receivers expect a pre-emphasis time constant = 75us (50us is a 'rest of the world thing')! - Additionally, availability of full, undistorted dynamic range will require a limiter -- Granting that pre-emphasis filters and prospective limiters are immaterial to your question -- still there it is...

Note also the non-standard power rail symbols (as noted on the schematic)...

As to my best guess as regards the mystery filters? -- First off - forget my musings Re: input biasing! Such is achieved via the non-inverting inputs...

So... I can but guess their (i.e. the 100ms "mystery filters' ") are with the function of canceling common-channel noise (or, perhaps, control of inter-channel gain disparity) at the expense of slightly reduced channel separation--- Embarrassingly, analog audio processing is not my 'strong suit' --- Please link the image with your question wherever you see fit that we might both learn something!

Best regards
HP

Last edited:

#### Hypatia's Protege

Joined Mar 1, 2015
3,226
@Aleph(0) -- Re AAC login difficulties...

Please rest assured that it's not personal...

I've just made several attempts (local RTC first set to my hub locale's correct CDT then ± 2 hours each 'side' thereof (with appropriate attention to the date [i.e. 1'st vs 2'nd]) -- with no luck! --- It seems the browser (FF) may be a 'factor' inasmuch as I was able to login only after 'spoofing' an x86 Pale Moon VM session (but still with issuance of my Genuine IP and my hardware's genuine Machine Access Code) --- Following which 'bare naked' Firefox sessions were allowed login but not file uploads... Only the VM session (through which this message is being composed/transmitted) is allowed full functionality at this time... --- Again! Inasmuch as we are experiencing the self-same difficulty --and-- Spoofing of neither my IP nor MAC are required for full functionality - no cause for entertainment of 'ideas of persecution' exists!

For the morbidly curious -- Here's the error screen 'in full living colour'..

Best regards and well wishes for better luck than I...
HP

Last edited:

#### Aleph(0)

Joined Mar 14, 2015
597
@Aleph(0) -- Re AAC login difficulties...

Please rest assured that it's not personal...

I've just made several attempts (local RTC first set to my hub locale's correct CDT then ± 2 hours each 'side' thereof (with appropriate attention to the date [i.e. 1'st vs 2'nd]) -- with no luck! --- It seems the browser (FF) may be a 'factor' inasmuch as I was able to login only after 'spoofing' an x86 Pale Moon VM session (but still with issuance of my Genuine ISP and my hardware's genuine Machine Access Code) --- Following which 'bare naked' Firefox sessions were allowed login but not file uploads... Only the VM session (through which this message is being composed/transmitted) is allowed full functionality at this time... --- Again! Inasmuch as we are experiencing the self-same difficulty --and-- Spoofing of neither my ISP nor MAC are required for full functionality - no cause for entertainment of 'ideas of persecution' exists!

For the morbidly curious -- Here's the error screen 'in full living colour'..

View attachment 185218

Best regards and well wishes for better luck than I...
HP
HP tnx for helping and proving I'm not only one effected by bug! So anyhow I'll make rest of my reply on this subject to feedback forum where I say it belongs anyway!

#### Aleph(0)

Joined Mar 14, 2015
597
HP abt networks in preamp stages of FM Stereo multiplexer that u circled here:

HP I don't follow explanation in your email saying purpose of networks are to make sure R=(R-L) and L=(L-R) b4 application to mux Cuz channels are ALREADY 100% SEPARATE at input! And anyhow I say it would cancel signals that are supposed to be common to both channels (like as example vocals on most recordings) so I say it seems more like karaoke machine than preamp!

HP Being serious I know I'm missing something! But I just can't get why there should be any interchannel signal sharing at all before time division multiplexing!

#### Hypatia's Protege

Joined Mar 1, 2015
3,226
HP I don't follow explanation in your email saying purpose of networks are to make sure R=(R-L) and L=(L-R) b4 application to mux Cuz channels are ALREADY 100% SEPARATE at input!
And anyhow I say it would cancel signals that are supposed to be common to both channels (like as example vocals on most recordings) so I say it seems more like karaoke machine than preamp!
Yebut we're not talking 100% cancellation! Moreover, I draw your attention to the fact that the 'cross-channel' networks are each with a TC = 100ms -- Hence my guess that said arrangement is with the purpose of low-frequency noise abatement and/or channel-to-channel gain equalization... But again! As previously stated I'm rather 'weak' on analog audio processing systems! -- I encourage you to post your inquiry to appropriate fora!

Tell ya what... Attached to this post is an LTspice 'sub circuit' model for the TL072 such that you may readily 'play with' the channel input arrangement in the simulator --or-- better still, you might take your own advice and merely breadboard it! -- In any event I maintain my suggestion that you seek insight from those having some degree of expertise with (analog) audio processing systems -- I apologize that --interesting though it is-- I am not at leisure to mentally 'wade through' this at present...

Very best regards
HP

#### Attachments

• 1.1 KB Views: 7

#### Aleph(0)

Joined Mar 14, 2015
597
Attached to this post is an LTspice 'sub circuit' model for the TL072 such that you may readily 'play with' the channel input arrangement in the simulator --or-- better still, you might take your own advice and merely breadboard it!
HP TNX! I'll do both!

In any event I maintain my suggestion that you seek insight from those having some degree of expertise with (analog) audio processing systems
HP I definitely want to post question to like _Analog and Mixed Signal_ or whatever is most appropriate forum but cuz of me being basically in _limbo_ finding acceptable VPN I'm never sure when I'll be back and there's no chat faux pas ruder than ingratitude of starting thread and _disappearing_ after OP!

#### joeyd999

Joined Jun 6, 2011
4,425
@Aleph(0) -- Re AAC login difficulties...

Please rest assured that it's not personal...

I've just made several attempts (local RTC first set to my hub locale's correct CDT then ± 2 hours each 'side' thereof (with appropriate attention to the date [i.e. 1'st vs 2'nd]) -- with no luck! --- It seems the browser (FF) may be a 'factor' inasmuch as I was able to login only after 'spoofing' an x86 Pale Moon VM session (but still with issuance of my Genuine IP and my hardware's genuine Machine Access Code) --- Following which 'bare naked' Firefox sessions were allowed login but not file uploads... Only the VM session (through which this message is being composed/transmitted) is allowed full functionality at this time... --- Again! Inasmuch as we are experiencing the self-same difficulty --and-- Spoofing of neither my IP nor MAC are required for full functionality - no cause for entertainment of 'ideas of persecution' exists!

For the morbidly curious -- Here's the error screen 'in full living colour'..

View attachment 185218

Best regards and well wishes for better luck than I...
HP
I think I found your problem:

#### Hypatia's Protege

Joined Mar 1, 2015
3,226