What is minimum plate supply impedance all about? The 5U4G tube specifies a minimum of 75 ohms per plate for a capacitor input supply. But I'm doing up a choke input supply with: TOROID (full wave) --> 5U4 --> 15H --> 40uf. I'm assuming I don't have to worry about this parameter when doing choke input, correct? The reason I ask is because my transformer is a toroid and I know those have less impedance than a vintage EI transformer. Is the minimum supply impedance restriction specified for safety or longevity of the tube? I remember as a kid we had a TV set where the 5U4 would literally explode, happened twice, then dad junked the TV.
My toroid has 3.5 ohms primary DCR (with coils in parallel for USA 115v). And it has secondary DCR of 80 ohms (40-0-40 ohms). 40 ohms is way below the 75 ohms per phase in the data sheet. BUT the data sheet only lists minimum supply impedance for capacitor input PS. Final question... Am I okay using this 40 ohm toroid for a choke input supply since the data sheet doesn't mention this parameter for that topology?
https://frank.pocnet.net/sheets/127/5/5U4G.pdf
My toroid has 3.5 ohms primary DCR (with coils in parallel for USA 115v). And it has secondary DCR of 80 ohms (40-0-40 ohms). 40 ohms is way below the 75 ohms per phase in the data sheet. BUT the data sheet only lists minimum supply impedance for capacitor input PS. Final question... Am I okay using this 40 ohm toroid for a choke input supply since the data sheet doesn't mention this parameter for that topology?
https://frank.pocnet.net/sheets/127/5/5U4G.pdf
Yes. Choke input is different.
According to 6A3Summer in another thread:
Example:
Take a power transformer, primary 120VAC DCR = 30 Ohms, secondary 360-0-360VAC, DCR of 1/2 secondary = 50 Ohms.
Then, 30 Ohms stepped up by the primary to secondary ratio of 3; 3 x 30 = 90 Ohms
50 Ohms of 1/2 secondary
90 + 50 = 140 Ohms.
That means to me, if you're worried about it, put 22R - 33R in series with each 5U4 plate.
I have used toroidal PT with choke input. it works as expected except for initial magnetizing current like all toroids (10's of amperes).
According to 6A3Summer in another thread:
Example:
Take a power transformer, primary 120VAC DCR = 30 Ohms, secondary 360-0-360VAC, DCR of 1/2 secondary = 50 Ohms.
Then, 30 Ohms stepped up by the primary to secondary ratio of 3; 3 x 30 = 90 Ohms
50 Ohms of 1/2 secondary
90 + 50 = 140 Ohms.
That means to me, if you're worried about it, put 22R - 33R in series with each 5U4 plate.
I have used toroidal PT with choke input. it works as expected except for initial magnetizing current like all toroids (10's of amperes).
You should be fairly safe using a 15H choke. The main concern is what they call hot switching current which you avoid with a choke input filter.
Rectifier tube life should be fairly good.
Rectifier tube life should be fairly good.
Minimum winding resistance and maximum first filter cap value are roundabout ways to stay within the peak repetitive current limit of the rectifier at maximum supply voltage and load current. PRR will be less with lower voltage or lower load current.
The cathode can only emit so much... but it's very hard to calculate, and few have the equipment to measure it. So simulate the circuit with PSUD. And allow some extra margin if you're not using old stock rectifiers - modern ones may not meet the old specs.
The cathode can only emit so much... but it's very hard to calculate, and few have the equipment to measure it. So simulate the circuit with PSUD. And allow some extra margin if you're not using old stock rectifiers - modern ones may not meet the old specs.
If you have a scope, measuring the peak current is a no brainer.
Install a 10R resister into the negative return lead to the power transformer.
The peak current is simply Ohms Law.............Ipeak = Epeak / 10.
Another method uses a peak responding VM. The peak volts indicated again, apply Ohms law.
But I think these daze most DVMs are RMS responding.
I put a 10R resister into the -ve return lead of all my builds, Real convenient, 🙂
Install a 10R resister into the negative return lead to the power transformer.
The peak current is simply Ohms Law.............Ipeak = Epeak / 10.
Another method uses a peak responding VM. The peak volts indicated again, apply Ohms law.
But I think these daze most DVMs are RMS responding.
I put a 10R resister into the -ve return lead of all my builds, Real convenient, 🙂
For reference, the Sylvania datasheet for 5U4G includes the equation used by 6A3Summer above. That equation is also used in the Source Impedance Calculator within PSUD2.
Another advantage of PSUD2 is that it estimates the two key rectifier current limit levels - steady state peak plate, and transient peak plate levels - for comparison to the datasheet limit levels for those parameters - so that you get a better appreciation of how closely those two peak values approach the limits for practical levels of choke inductance and the following filter capacitance.
Another advantage of PSUD2 is that it estimates the two key rectifier current limit levels - steady state peak plate, and transient peak plate levels - for comparison to the datasheet limit levels for those parameters - so that you get a better appreciation of how closely those two peak values approach the limits for practical levels of choke inductance and the following filter capacitance.
Inductive reactance is 2(PI)FL.
As long as the 15 H choke does not saturate, you are looking at 5.5K resistance as far as what the rectifier sees. DC resistance does not matter so much in this case.
Also you could increase the capacitor a bit without breaking things.
Maybe run PSUD to double check for grins.
As long as the 15 H choke does not saturate, you are looking at 5.5K resistance as far as what the rectifier sees. DC resistance does not matter so much in this case.
Also you could increase the capacitor a bit without breaking things.
Maybe run PSUD to double check for grins.
The rule for choke input filter requires that you use a choke that has at least the Minimum Critical Inductance.
60Hz power mains, 120Hz full wave rectification:
Minimum Critical Inductance = 350 / mA DC load.
Example: DC Load is 100mA
350 / 100 = 3.5 Henry
Use at least a 3.5 Henry choke.
Want some extra margin, in case the load current is a little higher, use a 5 Henry choke.
50Hz power mains, 100Hz full wave rectification:
Minimum Critical Inductance = 420 / mA DC load.
Example: DC Load is 100mA
420 / 100 = 4.2 Henry
Use at least a 4.2 Henry choke.
Want some extra margin, in case the load current is a little higher, use a 5 Henry choke.
Using any choke that has less than the critical inductance . . .
Then the power supply does not have a true choke input filter.
In order to get a little more B+ voltage, I sometimes use a small capacitance before the choke,
between 1uF and 4uF as needed to get the required voltage.
Using a 4uF input filter cap is a much easier load on a tube rectifier than a 60uF input cap is.
This modified circuit is somewhere between the characteristics of a true cap input filter, and the characteristics of a true choke input filter.
60Hz power mains, 120Hz full wave rectification:
Minimum Critical Inductance = 350 / mA DC load.
Example: DC Load is 100mA
350 / 100 = 3.5 Henry
Use at least a 3.5 Henry choke.
Want some extra margin, in case the load current is a little higher, use a 5 Henry choke.
50Hz power mains, 100Hz full wave rectification:
Minimum Critical Inductance = 420 / mA DC load.
Example: DC Load is 100mA
420 / 100 = 4.2 Henry
Use at least a 4.2 Henry choke.
Want some extra margin, in case the load current is a little higher, use a 5 Henry choke.
Using any choke that has less than the critical inductance . . .
Then the power supply does not have a true choke input filter.
In order to get a little more B+ voltage, I sometimes use a small capacitance before the choke,
between 1uF and 4uF as needed to get the required voltage.
Using a 4uF input filter cap is a much easier load on a tube rectifier than a 60uF input cap is.
This modified circuit is somewhere between the characteristics of a true cap input filter, and the characteristics of a true choke input filter.
Thanks, this PS is going to be for the Salas 6V6 line Amp project, before I incorporate the shunt regulator I want to build it with a traditional choke in supply but leave room in chassis for the shunt board and heatsink some day. Salas specified 100ma minimum choke and each channel draws 22ma or so. I have a 15H 100ma choke, which may be overkill but the DCR of this choke and my current T1 is landing me pretty close to the 340V when loaded to 50ma using 5U4 and following LC stages.
I was not aware that I could put a small capacitor up front and still have it behave like a choke input! A good use for an MKP DC link capacitor there, and a good way to get a little more voltage to overcome the drop out of the regulator once I add that piece some day, without having to change out T1! Nice, thanks, knowing I have wiggle room upwards using small capacitor will make fitting a regulator later easier and let me land it at 340V for now without the regulator.
Note that a small value cap before the choke will carry a high AC voltage, so you need to confirm that such a cap has a rating to cover the ACV at twice mains frequency. The cap negative leg should also return as directly to the transformer winding as practical to minimise the current loop area. The resulting B+ level is quite sensitive to cap value, so you may need to be lucky if you only have one or two cap values at hand and want to get close to a target B+ voltage.
Okay, right now the breadboard gives me 345 volts with 50ma load, the Salas 6v6 will probably draw less and that project specifies 340 B+. I have no resistors in the train only chokes. When I add a shunt regulator later Ill need more volts to overcome its loss and the drop out plus some. Without changing T1 would be great. I'll experiment with a 0.5 uf 900v MKP and try other values up to 4.7 uf to see the rate voltage increases, but I don't want it to become a capacitor input supply but ok if its "mostly" a choke input supply. Salas does use a 4.7 uf input capacitor I see. Is that small enough to still be "somewhat" a choke input PS? This is new to me that there might be a range of small capacitance where the supply is neither fully of one topology or the other. This seems like a better way to "trim" a power supply instead of trimming downward with resistor or choke DCR. Just hit your voltage as choke input topology and add a little bit of capacitor to trim upwards to your target voltage at load instead of downwards with a resistor, no?.
https://www.diyaudio.com/community/threads/6v6-line-preamp.102352/page-5#post-1598914
Last edited:
Adding input capacitance 'slides' the B+ level from nominal 0.9x (choke input) to 1.4x (cap input) the transformer winding voltage - so quite a substantial range, and hence the comment about being quite sensitive. I'd suggest 4u7 would be quite close to full cap input. PSUD2 is a good tool to get a feel for the likely B+ voltage expected when changing the input cap value.
A '900V MKP' is imho too ill-defined to ensure it is a suitable capacitor for this application - it is best to chase down the manufacturer datasheet.
A '900V MKP' is imho too ill-defined to ensure it is a suitable capacitor for this application - it is best to chase down the manufacturer datasheet.
It is sensitive but manageable in sub uf increments 0.5 uf slides it up from 345 (no capacitor) to 391 volts. 40 uf (5U4 limit) slides it up from 345 (no cap) to 542 volts!
0.22 uf slides it from 345 to 353, so its still "mostly" choke input and that range is nice for trimming to the regulators needs. 4.7 uf slides it from 345 to 537 so yes Salas has very much a capacitor input supply there! I was worried about adding the regulator later and not having enough voltage. Maybe a high voltage 0.22 or 0.33 uf film cap can easily trim up the additional few volts needed later? Or two 630 volt .47 uf orange drops in series to up the voltage rating. This whole idea of "trimming upward" is something I never thought of doing. Usually when people are short on voltage they are looking for a new T1 or shedding resistance, but I have no resistors its all LCLC with a tail L/R pair of LC's.
0.22 uf slides it from 345 to 353, so its still "mostly" choke input and that range is nice for trimming to the regulators needs. 4.7 uf slides it from 345 to 537 so yes Salas has very much a capacitor input supply there! I was worried about adding the regulator later and not having enough voltage. Maybe a high voltage 0.22 or 0.33 uf film cap can easily trim up the additional few volts needed later? Or two 630 volt .47 uf orange drops in series to up the voltage rating. This whole idea of "trimming upward" is something I never thought of doing. Usually when people are short on voltage they are looking for a new T1 or shedding resistance, but I have no resistors its all LCLC with a tail L/R pair of LC's.
Last edited:
Your post got me thinking of this concept of trimming upward, very cool to take advantage of micro-sized capacitance up front. Hybrid choke input.
Last edited:
Capacitor input filter:
At 120 Hz full wave rectification, a 10 uF capacitor capacitive reactance, Xc, is 133 Ohms.
The rectifier tube has to drive that impedance.
With 50mA DC load, the approximate peak to peak ripple would be 0.050 x 133 Ohms = 6.6 V; the middle of that ripple is 3.3V.
Neglecting the tube rectifier drop, the output would be about 3V less than 1.4 x the secondary rms
(full secondary for bridge, or 1/2 secondary for center tap and full wave operation).
In Between Cap input filter and Choke input filter:
At 120 Hz full wave rectification, a 0.5 uF capacitor capacitive reactance, Xc, is 2660 Ohms.
The rectifier tube only has to drive that impedance.
With 50mA load, the approximate peak to peak ripple would be 0.050 x 2653 Ohms = 133 V, the middle of that is 66V.
Neglecting the tube rectifier drop, the output would be about 66V less than 1.4 x the secondary rms
(full secondary for bridge, or 1/2 secondary for center tap and full wave operation).
The small input cap and the choke that follows may resonate at the full wave rectified frequency.
fo = 1 / (2 x pi (Root L x C))
A 5H choke and 0.4uF cap resonates at 113Hz, too close to 120Hz full wave rectification.
At 120 Hz full wave rectification, a 10 uF capacitor capacitive reactance, Xc, is 133 Ohms.
The rectifier tube has to drive that impedance.
With 50mA DC load, the approximate peak to peak ripple would be 0.050 x 133 Ohms = 6.6 V; the middle of that ripple is 3.3V.
Neglecting the tube rectifier drop, the output would be about 3V less than 1.4 x the secondary rms
(full secondary for bridge, or 1/2 secondary for center tap and full wave operation).
In Between Cap input filter and Choke input filter:
At 120 Hz full wave rectification, a 0.5 uF capacitor capacitive reactance, Xc, is 2660 Ohms.
The rectifier tube only has to drive that impedance.
With 50mA load, the approximate peak to peak ripple would be 0.050 x 2653 Ohms = 133 V, the middle of that is 66V.
Neglecting the tube rectifier drop, the output would be about 66V less than 1.4 x the secondary rms
(full secondary for bridge, or 1/2 secondary for center tap and full wave operation).
The small input cap and the choke that follows may resonate at the full wave rectified frequency.
fo = 1 / (2 x pi (Root L x C))
A 5H choke and 0.4uF cap resonates at 113Hz, too close to 120Hz full wave rectification.
Might be an interesting variation to include a defined resistor in series with the input cap, something much larger than ESR, and approaching a wastefully low-Z snubber. Something natively including the twitchy voltage tuning and power transformer secondary resonance damping. And, would that make it less or more twitchy?
All good fortune,
Chris
All good fortune,
Chris
This was going to be my next question. So a resistor and smallish cap in series instead of just a cap. A "snubber input" PS to stop oscillation and up the voltage both? This is something for the breadboard to see if it's oscillating on the peaks or not?
A 5 Henry choke has 3,770 Ohms of inductive reactance at 120Hz.
Suppose the cap value you picked (to get a specific B+ Volts) resonates with the 5 Henry choke at 120Hz.
That means the capacitor also has 3,770 Ohms of capacitive reactance at 120Hz.
So if you put a 377 Ohm resistor in series with the capacitor, the Q of the LC resonator is 10.
That is a fairly high Q, and the voltage may still rise significantly.
And, that 377 Ohm resistor is going to dissipate power.
You do not have to use the 377 Ohm resistor in series with the capacitor . . .
Instead, you can change the 5H choke to a 10H choke
(yes, more $$$, more weight, and more space).
But doubling the choke Henrys lowers the resonance to 70.7% of the original resonant frequency.
10 Henry and the same capacitor: 120Hz x 0.707 = 84.8 Hz resonance, right in the middle between 120 Hz and 60 Hz.
OK.
Suppose the cap value you picked (to get a specific B+ Volts) resonates with the 5 Henry choke at 120Hz.
That means the capacitor also has 3,770 Ohms of capacitive reactance at 120Hz.
So if you put a 377 Ohm resistor in series with the capacitor, the Q of the LC resonator is 10.
That is a fairly high Q, and the voltage may still rise significantly.
And, that 377 Ohm resistor is going to dissipate power.
You do not have to use the 377 Ohm resistor in series with the capacitor . . .
Instead, you can change the 5H choke to a 10H choke
(yes, more $$$, more weight, and more space).
But doubling the choke Henrys lowers the resonance to 70.7% of the original resonant frequency.
10 Henry and the same capacitor: 120Hz x 0.707 = 84.8 Hz resonance, right in the middle between 120 Hz and 60 Hz.
OK.
The DCR of a 15H 100mA choke could be more than 200 ohm. The C of any LC resonance will be lower due to the DC load side capacitance. The diodes are effectively on all the time (if filter input C is low enough) so the low value input C is being dominantly driven/controlled by the transformer secondary voltage, not by an LCR self-resonance.
My 15H choke is 256 DCR. That is the first choke I'm committed to, I already bought an enclosure to pot it and for cosmetics as its on chassis top. The train is LCLC then splits left/right to dedicated channel LC's for each channel (lessen crosstalk). No electrolytics all DC link (I had them), no resistors. My final voltage is spot on 340-345 volts unregulated. As mentioned later on I'll need 20-30 more volts or so as each channel will have its own shunt regulator board. PSUD2 says adding a C1 of 0.22 or 0.33 uf gives me that extra 20-30 volts, easy-peasy and cheaply. I just need to learn more about reactance, oscillation, etc. to see if this is stable, just throwing in a 0.22 cap, maybe its too good to be true. Right now I'm kinda lost without filling some big gaps in my AC theory.