I'm working on a small AB amp and getting a lot more sag in the B+ than I expected. It's a pair of ECL82s running 35mA idle including screens (combined total for the pair, not each tube), a few more mA for the preamp. My power transformer is a Hammond 269EX rated for 75mA. I'm using a full wave rectifier so DC current should be the same as AC. After 100uF of filtering DC at idle is 265. My plate to plate load is 12k8. In simulation, excluding the preamp, at full drive the average current draw (Again, including screens) should be just about 75mA so I'm a bit over rating all in but not by too much. When I measure the current accross a 1 ohm sense resistor on the cathodes my multi-meter is seeing about 125mA of current (Not sure if this is a proper average or a peak) and B+ drops to 210. I checked the screens and they go from 3mA to 8 with 1k grid stoppers from idle to full output.
- Is a 1 ohm sense on the cathodes a good way to measure dynamic current draw in this situation? I'm using an Agilent 34401A.
- If it is, something must be wrong as I would think peak current would be closer to 80mA and average much lower. (265v/(.25(12k8))
- Generally speaking is there a good way to estimate/calculate the "correct" current rating for your power supply assuming sag of more than 10% is undesirable?
Let's look at the databook.
http://www.mif.pg.gda.pl/homepages/frank/sheets/030/e/ECL82.pdf
With 250V 10K load and *Sine* output we should have 87mA total cathode current.
Adjusting for 12.8K and 265V or 210V I get 72 and 57mA.
What strikes me is that your reading is "twice" what it should be. Are you very sure you do not have 8 ohm load on 16 ohm tap?
http://www.mif.pg.gda.pl/homepages/frank/sheets/030/e/ECL82.pdf
With 250V 10K load and *Sine* output we should have 87mA total cathode current.
Adjusting for 12.8K and 265V or 210V I get 72 and 57mA.
What strikes me is that your reading is "twice" what it should be. Are you very sure you do not have 8 ohm load on 16 ohm tap?
Attachments
on tube amps, no load B+ and loaded B+ can be far away, so if your unloaded B+ is say 350v, expect to see that way below 300 volts when in use...more if there are more resistance along the B+ line in series like a choke or a power resistor...
http://www.hammondmfg.com/pdf/EDB269EX.pdf
with capacitor input filtering, idle B+ is 301v dc unloaded... with tubes warmed up you report 265 vdc and that is a reasonable reading....
i do not see what your issue is...is the amp working ok?
http://www.hammondmfg.com/pdf/EDB269EX.pdf
with capacitor input filtering, idle B+ is 301v dc unloaded... with tubes warmed up you report 265 vdc and that is a reasonable reading....
i do not see what your issue is...is the amp working ok?
Your expectations are unrealistic and doubly so for such an underrated transformer.
10% sag is way too low and can only be achieved with a heavily overrated PT (2x to 3X above what´s *actually* needed), abbundant filtering,etc.
Not your case.
20% or so might be more realistic.
You are starving that amp.
You need *¨*at least* 1.5X the expected current consumption, 2X being more realistic.
Wrong.
That woukd be approxuimately right foir purely resistive loads ... and nowhere else, specially on a capacitive input supply.
Capacitors are charged in narrow peaks which are way higher than *average* current.
Instantaneous voltage drop is of course proporcional to instantaneous current and thermal loss/dissipation is proportional to instantaneous current *squared. .
Hint: current bursts have way higher RMS value than same average valkue sinewaves.
And RMS defines thermal loss.
Not sure about the huge difference between measured current ans simulation, one or both must contain some kind of error.
In principle I trust the Multimeter ("measuring rulez"
) and suspect the parameters you introduced into your simulation.125mA average from a 75mA rated PT is a sure fire way to kill it.
yes
Quite wrong assumption, it does not work that way.
Use some kind of Power Supply Designer software, but 10% regulation is a very high goal to achieve, not within your budget constraints anyway.
Brian,
Not sure if you are simulating the power supply, but PSUD2 software is a good crosscheck and forces you to insert the transformer winding resistances.
Your meter may be able to measure pk current going to common screen supply node, or in to a screen stopper (I recall that meter have great signal/maths functions). The peak level should increase from idle to full load and occurs at the minimum of the anode voltage swing.
Do you have cathode bias, or fixed - as you don't clarify that. At idle, the sensed cathode current is DC. It gets trickier to interpret for large output signal conditions - as you could be measuring proper rms (if the waveform) crest factor is ok for the meter, but you may want to also measure peak - as that relates to OPT winding voltage drop.
Measurements get uglier to interpret if you are clipping the plate voltage waveform - you may have harmonic maths info with the meter if you don't have a scope or spectrum analyser - that will tell you when clipping starts.
Last edited:
The issue is two fold: Firstly, I want to know where i'm going wrong with my estimates; second, I'm getting the scariest crossover distortion in overdrive (Way more than I've seen from charging up coupling caps) and I was thinking that the screen voltage dropping 20% would be a contributor.
Well, ya, but I didn't think it was underrated until I made this thread.
Going forward I'll take the full-power datasheet numbers and double them.
I was communicating the ratings and that my 75mA rating was DC value, not AC prior to rectification. I reference this Hammond design guide (fourth one down on the left is relevant): http://www.hammondmfg.com/pdf/5c007.pdf
Not sure about the huge difference between measured current ans simulation, one or both must contain some kind of error.
In principle I trust the Multimeter ("measuring rulez"
Well, for one thing the screens in spice don't draw any current at idle... that's not helping
Don't want to do that!
Good, thanks
This formula tells me where the load line meets the y axis of the curves (Close to, but not exactly, where the 0v curve is) but I see now that more margin is needed.
I didn't have the regulation detail for the Hammond transformer to simulate. I communicated no budget constraints and fortunately I don't really have any. I started the project with what was on hand and I'm comfortable picking up a new power transformer to match my needs.
Fixed bias, about -30 to the grids via 220k resistors.
So I've got a Hammond H300519 I can liberate from another project. It's rated for 100mA DC which still seems insufficient for this little project but I can wire it up and see how it performs. It should idle around 350vDC so I'll have to up my impedance.
Yes; but the worst should happen is 1/2 primary (3,200 Ohms) with full 210V across it, all the time (utter square waves). Which is 65mA on my slip-stick.
Yes, grid crossover distortion is ugly. Learn to love it, or add large stopper resistors. Debiasing should vanish when stopper is about half of gridleak. Nobody ever goes that far, for several reasons. But 50K is a fine trial value to guide further fiddling.
how do you know it is wrong?
crossover distortion is a function if idle current bias,
as you look at the sine wave on the scope, 1 watt into an 8 ohm dummy load, you increase the idle current till the transition at the zero crossing is very smooth...if you follow recommended operating points in tube datasheets, then you have less chance of crossovers..
yes, there are two ways you can change the idle cathode current, first is by decreasing the g1 negative voltage; i.e from -5v to say -4 increase cathode current and vice versa....
and second by the g2 voltage, increasing the voltage at g2 increases plate current and vice versa...
and why overdrive? is this a guitar amp?
perhaps you mean to say maximum input signal plate current...where zero signal and maximum signal plate currents are much different, look at the datasheets to see this...
This is the guitar amps forum so that is probably the case.
One thing many people don't realize that a class A amp or even a class AB amp loses its class distinction when you plug in that stomp box and set it's controls for the heart of the sun!
The output tubes become on-off switches (some are better than others) and peak voltages can rise to well over the usual twice B+ levels, especially when driven hard at the speakers resonant frequency which is often in the guitar's frequency range.
Strong signal crossover distortion is usually caused by bias shift. It does not always come from the output tubes. I found out an an early age that it is possible to generate blocking distortion in the input stage....just plug a Fuzzface into an 5C1 Champ, turn all the knobs up full and thrash away. The amp will blast out a loud sound for a second or two, then go silent for a while and slowly recover to a very distorted sound. I found that this could be cured by connecting my voltmeter (a plastic VOM from Lafayette) on the grid of the input tube. The 5C1 uses grid leak bias on a 6SJ7 which really doesn't like being slammed with 4 volts of signal.
When working on a prototype amp guitar or HiFi, I use a scope across a 1 ohm resistor in the output tube's cathode to determine the continuous and peak current and the true operating class of the amp at various power levels. Many voltmeters can't display an accurate value when the current is not continuous, even some "true RMS" meters.
Some scopes have a math function that allows the subtraction of one channel from another. You can put a probe on either side of a stopper resistor to measure grid (G1 or G2) current if this function is available.
There are those who say that an amp is NOT a class A (or whatever class) amp if if ever leaves class A under normal operating conditions. If we follow these strict guidelines, there are NO class A guitar amps. Truth is that almost all amplifiers have a power range where they run in class A, for some this is small, say zero to half a watt. For others it may be much larger. This is a function of load impedance and bias current.
I have not used that particular tube so I don't know how it's screen current behaves, but the screen current on most tubes skyrockets when the tube's plate is pulled down below the screen grids potential. This can fry tubes in a guitar amp that will be driven hard, and some means of limiting the grid current must be employed.
Hint:
Look inside the tube in a darkened room and thrash away at full crank. Is something other than the cathode glowing? Does the tube appear to gain brightness when playing hard, then dim at idle? If so, they are not going to live long, because it's probably the screen grid providing the light show. Poorly assembled EL34's are really bad at this.
True, I would say almost every guitar player (except for people who play nothing but jazz) want at least some distortion out of their guitar amp.
I think one of the things which is overlooked is that the usual electromagnetic guitar pickup itself produces a very distorted waveform. What comes out of the pickup is nothing like the actual physical movement of the string. After the initial big transient, the string movement seems to be almost sinusoidal (with a little bit of 2nd harmonic.) But what comes out of the pickup is usually very different. This is because the magnetic field around the pickup poles does not decrease linearly with distance - and this causes the induced voltage in the pickup to be distorted.
This pickup distortion seems to be a major reason why electric guitars tend to sound harsh compared to acoustic guitars. And, oddly enough, by having some of the right type of distortion in the guitar amp, often the guitar itself ends up sounding less harsh - good guitar "clean tone" requires some distortion from the guitar amp.
Of course, many guitarists actually want their guitars to sound harsh and aggressive, and in that case, massive amounts of distortion are expected from the guitar amp.
-Gnobuddy
One half the OT (centre tap to one end) has one-quarter of the end to end impedance, no? So I think, worst case, that should be 1600 ohms and 210 volts, for about 130 mA. That's with a well-behaved resistive load, things might be worse during transients with an actual loudspeaker (which generates its own EMF which gets pumped back through the transformer.)
I found that reducing the maximum output signal amplitude from the phase splitter also helped. Once grid current starts flowing, the grid caps can charge up to a voltage that depends on both the ratio of (R_grid_stopper/R_leak), and also on the peak signal voltage coming from the phase splitter. So reducing the latter can help reduce blocking distortion.
A couple of years ago I built a little 6AK6 amp that had a MOSFET "source-o-dyne" capable of spitting out about 75Vpp at each output, while the output valves were biased at (-10 volts) quiescent. Once you drove it hard enough for grid current to start flowing in the 6AK6s, if you had no grid stoppers, the coupling caps would be trying to charge up to (-37.5 V) - nearly four times the quiescent bias!
Starting with typical grid stoppers of a few kilo ohms, and not understanding this at the time, of course I had all sorts of problems with blocking distortion. In the end, using 220k grid stoppers on the 6AK6s made for a huge improvement. Reducing maximum drive signal to the source-o-dyne from the preceding stage also helped.
Incidentally, the very small input capacitance of the 6AK6s doesn't cause any audible treble rolloff with a 220k grid stopper. I didn't go bigger only because of the datasheet restrictions on maximum grid bias resistance.
-Gnobuddy
Attachments
Last edited:
Maybe Im measuring wrong? I'm leaving my meter in DCv (when I check bias current) and measuring across the 1 ohm cathode resistor (And getting 125mV). Should be using ACv? I tried briefly yesterday and got 40mV (need to double check that)... if I added 40mA to the bias current for the two tubes I'd already be past the 75mA rating.
I've actually upped my grid stoppers now (from 2k5 to 68k) because I was getting some weird linearities in the phase inverter when the output tube grids start pulling current. IME big grid stoppers do help but only much. The ratio of drive signal to bias voltage is just too big on these small amps and you need other tricks.
Before we all jump too far into a side discussion on crossover distortion (I really, really need to settle on what my new power transformer will be rated to handle) I will say that this project is generation three in pursuit of a small-amp-with-big-amp-tone.
I started with ECC99 triodes and didn't like the tone, then spent a month fiddling with ECC99's in cascode (hoping to simulate pentode tone) and things were different but less pleasing (As well as very inefficient), so now I've moved on to little power pentodes.... I skipped 6AK6 due to socket type, ECL80 due to the shared cathode, and jumped at ECL82 because there is lots prior art to reference (and, I thought, I had transformers that would fit the bill). The Watkins Dominator series even had a version that ran ECL82 that I could copy. Frankly the amp is so loud I might as well be using EL84s.
Small "power" tube sections tend to have very low bias voltage requirements; low bias means very early clipping of the positive part of the waveform which leads to aggressive charging of the coupling caps. I believe this is why 18W Marshall's are notorious for crossover while their bigger brothers are not (Of course it's there, but less drastic). I've run though lots of the known tricks (short of DC coupling with mosfets) but I'm going to start another thread later once I get the PS stabilized.
Brian
The bias current is DC, so that makes sense. But once you start driving the amp with a signal, you have an AC signal, riding on top of a DC (bias).
A true-RMS meter might make some sense of this, but you're probably more interested in the peaks of that AC signal than in the RMS value. And to see this, you really need an oscilloscope.
I experienced exactly this problem (and referred to it a couple of posts ago.) Consider attenuating either the signal from PI to output valves, or the signal entering the PI, so it is no longer capable of overdriving the output stage quite so deeply. If the output valves only need 10 Vpp to begin to clip, there isn't much point driving them with a 100 Vpp signal...a square wave is pretty much a square wave!
Yeah, that's also exactly what I experienced too. Depending on the speaker you use, a single watt is as loud or louder than a trumpet blast - and how many contemporary households will allow for trumpet-blasts in the living room?The Watkins Dominator series even had a version that ran ECL82 that I could copy. Frankly the amp is so loud I might as well be using EL84s.
Going at it from the SPL end rather than from the watts end, you can easily calculate that you only need a few milliwatts (maybe a few tens of milliwatts) of power to the speaker to reach the maximum household-friendly SPL. There are plenty of online SPL tables to give you a reference point - I settled on a household vacuum cleaner as the loudest sound the neighbours might be willing to put up with, and that's typically "only" around 70 - 75 dB SPL.
Here is an interesting SPL table that also includes informed comments on how much your ears might be damaged by exposure to them: Everyday Sounds & Their Damaging Decibel Levels | Hearing and Balance Doctors
When I tried lowering amp power to milliwatts, I found the results really disappointing. The timbre (especially for clean tones) just disappeared - my ears simply weren't hearing the full spectrum any more (Fletcher-Munson), and those lovely subtle valve-induced harmonics might as well have never existed. And clean tones at milliwatt levels sounded too quiet, while distorted tones at milliwatt level sounded annoyingly loud.
In the end, I went back to a couple of watts in the power amp, combined with a preamp that had small-signal pentodes in it to generate all the distortion, and a master volume in between. This helped a lot, since I could now get usable clean tones as well as overdrive, without ever exceeding a reasonable SPL.
On a hunch, I sat down on the couch, pulled one of those little TV tray tables right up in front of me, and put the speaker on it, right in front of my face, and no more than two feet from my nose. This really helped: my ears were in the speakers near-field region, and I could get enough SPL there for my hearing mechanism to be able to function properly. At the same time, if you moved even three metres / ten feet away, there was very little SPL to disturb anyone else.
-Gnobuddy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.