Constant Current Supply as tail for valve Long Tail Pair?

[Diabolical Artificer]

Just built one of these and am puzzled. AFAIK a CCS using a NPN transistor is a current sink.
So how does that work SUPPLYING a cathode with current. Again AFAIK current flows off the cathode to the anode in a valve,

This has been bugging me for a while. It obviously works but how, if the current flow is up in a valve (as per usual schematic) but the current through a CCS is going the opposite way?

Andy.

[ukcol]

I think you may be confusing conventional current with electron flow. We construct models of how circuits work and sometimes it is convenient to think of so called "conventional current" which flows from positive to negative and at other times we work with electron flow which is negative to positive.

In the real physical world, of course, the actual flow of electrons is negative to positive. A good example is in the understanding of transistors we often talk about the flow of "holes" but in fact it is the electron movement (in the opposite direction) that creates the hole flow.

[julie_m]

Electrons flow from cathode to anode.

"Conventional current" flows in the opposite direction to electrons.

This is a historical accident; the assumed direction of flow was chosen before anyone had proved the existence of the charged particles responsible for carrying it. And by the time someone did, and found out the initial assumption had been wrong, it was too late to change the existing body of literature. So we just learned to with it, like we learned to live with having a prefix in the name of a fundamental unit .....

[ukcol]

Another point that may be helpful is to understand that a constant current source circuit is not the source of the current; the true source of current is the power supply, the CCS simply controls the current available so that it is constant.

[Silicon]

The arrowheads in diode and transistor symbols conform to the 'conventional' current flow direction and not the electron flow direction.

[Ambientnoise]

Of course a constant current "supply/source/sink " is a two terminal device so you can consider it to be a source or sink depending on your viewpoint.

Ken

[Radio Wrangler]

A theoretical constant current source is an energy source. If you leave an ideal one lying around on your bench not connected to anything it will start creating infinite voltage and cause no end of trouble.

What we're talking about here is an energy sink with a current limiting characteristic.

If we talk about conventional current, it enters the pair of valves at their anodes and leaves them at their cathodes. Both flows add together and pass into the constant current device. The constant current device lets the cathode voltage float as high as it needs to be in order to get the total current down to the value the constant current device wants.

David

[bikerhifinut]

thank you David for that succinct explanation of what the CCS does. I kind of knew what it did (I've used them and calculated resistor values) but never could explain properly what it did.
Sorry to Andy for hijacking the topic there.

The other Andy.

[Diabolical Artificer]

Thanks all, I thought it might be a conventional/electron mismatch.

No worry's Andy, think were both a bit sketchy on silicon. I'm using the circuit from here - http://www.pastisch.se/faktiskt/diyAudio-CCS-beta3.pdf . The HV one using a BC459 and MJE340. AFAIK Q1's base is set by the LED, a zener can be used instead which I think gives a better OP Z. Q1's emitter and the current of the CCS is set by R2 and 3. Q1 controls Q2 which is there to handle the extra voltage. R1 sets the current for the two LED's/voltage references. Are these two Q's in grounded base or common emitter config?

I've read quite a bit on CCS's including CCS Circard's which David recommended but admit to still being puzzled by some of the circuit's used, see pics attached. How does D work in pic one? And I have no idea how the one in pic 2 works. There's no voltage reference as such unless R15 is it. I couldn't get this circuit to work, the base wouldn't bias correctly, it sat at a much higher voltage.

I'm having a bit of trouble with the one I've built not having matched OP's. I'm using 36K matched anode resistors and a matched ECC82/12AU7. Grid one has a 1M to ground, g2 is grounded straight to ground and the CCS is supplied from - 70v. Bias is set to -6v ,HT is 350v .OP of anode's are direct coupled to a pair of cathode follower's. They should be bang on.

I'm getting an OP of 24.7v RMS on V1 and 24.5v RMS on V2 which seem pretty damn close but on the scope - P-P the mismatch is significant. In the past with a similar setup I've had them with no noticeable difference on the scope. Could it be the 1M R on g1? Both anodes are at 179 ( 167 on load line).

Andy.

[Diabolical Artificer]

Guess who forgot to make sure the "cal" knob on one channel was properly set? Doh! I'd also put a diode the wrong way round on the KF grid, this didn't help. LTP PS working like a good un.

Still interested in CCS's though.

A.

[wd40addict]

Re picture 2:

The 75v is divided down by R21/R16 to give approx 57v across R16.

57-0.6 = 56.4v across R15. Therefore current is set to 56.4v/2k8=20mA

(All voltages quoted with respect to the -75v line not 0!)

[ionburn]

Perhaps going off a little at a tangent but I like to fall back on Kirchhoff's, Thevenin and Norton's laws / theroms to understand what is going on in a circuit. That with the fact that if a transistor is forward biased it will conduct.

These things really helped me understand what is going on in a circuit, especially Kirchhoff's for flow of current (like ''rivers'). I visualise these things in my mind!

[Radio Wrangler]

Next installment:

Kirchoff happens at the junction node of the cathodes and the constant current source.

The constant current source allows the cathode voltage to float up until the valve takes the intended total current.

If the grid voltages shift, then the current source allows the cathode voltage to shift tracking them to keep the total current in the valves constant.

So the sum of the anode currents remains constant. The difference in grid voltages determines how that constant current gets shared between the two anodes.

The constant current means that the mean voltage drop in the two anode resistors is constant. The difference in anode currents sets the difference in the anode voltages.

We have a differential amplifier. The difference in grid voltages creates a difference in anode voltages and we have gain. The mean grid voltage may vary but it's the constant current device which sets the mean anode voltage. Our circuit has differential voltage gain, but blocks common-mode signals.

David

[Diabolical Artificer]

I've struggled with Kirchoff's law in the past, all the explanation's I've read left me with a fuzzy feeling between the ears. Then I found this bloke - https://www.youtube.com/watch?v=omSrIbJLk9c he speaks in a laid back voice like the HHGTTG book (Peter Jones) which helps to keep in check the feeling of panic and confusion that sometimes happens when info input get's too much. : )

I really need to sit down and practice though, he makes it look easy. He says treat every transistor as if it's in linear mode, or should I say solve it as if in linear mode, then that'll tell you if the Q is in cutoff or saturation. I may be getting a bit mixed up here with nodes or something else.

The problem I have with analysing circuit's is that in a real circuit there are more valves or Q's than given in some of the examples I've seen. Practice again would help.

Thanks for the explanations chaps.

Andy.

[Diabolical Artificer]

The CCS I knocked up, is this one - see attached. The LTP PS is an ECC82, HT 350v, 36k anode resistor's triode 1 has a 1M resistor on g1 to ground, triode 2 has g1 straight to ground, so the the grids are sitting at 0v, or should be, though normally V2 g1 is 0.03v. Cathodes connected together.The CCS has return connected to ground, input to -78v.

It has worked in the past, or I thought it did. At power up the LTP PS read - 76v, then dropped to -6v as the heaters warmed up. I was reading the voltage with a cheap DMM, which has shown some odd readings in the past, so ignored the minus sign as the PS seemed to be working. However after I tidied up the layout and tried to use a 6v8 zener in place of LED2 to get a higher Z, I got odd readings. The cathode rose to 200v with 300v on one anode and 40v on the other!. So I went back to using the LED instead of the zener.

Now the cathode's measure -10v despite twiddling the trim pot and anode's are at 48v and 120v. To figure out what the dicken's was happening I took voltage readings on the grids, the grounded grid is at zero but the one with the resistor is at + 10v! So I put a 10r resistor in series with the CCS and found the
CCS was sinking only 4mA .

Checked the CCS, everything reads ok, and rechecked my calculation's. R1 is 12k (12k nearest 78/0.005 =15,600 ohm's), R3 is a 100 ohm trimmer, R2 is 56ohm, so for a current of 0.011mA (5.5mA per triode) Re should be 90ohm's. Tested it with an LED and 10r resistor, measured 0.110v across the 10r so therefore 11mA, which is what it should be. So I'm struggling to figure out what is going on here. As I understand it the cathode should read +6v which is the bias point. Once warm the cathode passes current and our -76v should turn into +6v or whatever, a bit like when current flows through a resistor, we get a voltage drop.

I'm taking voltage reading's with the black lead to ground. Could it be that triode 2's grid is going positive and conducting so we get 10v because of the 1M resistor? But the anode's arn't glowing or the valve getting hot. Reading up on this for a possible explanation I found this - " this circuit may not work if Re is less than about 400 ohms because Vbe increases with current. If you need to sink more than about 2mA it is better to raise the base voltage by using an LED with a higher forward voltage, or zener diode." Got this from here - http://www.valvewizard.co.uk/ccs.html 4th to last paragraph.

No idea how to analize this circuit, I struggle when I have a negative voltage supply. For instance is the -78v rail now ground and ground now + 78v and HT =428v? Any help much appreciated.

Andy.

[wd40addict]

There is nothing attached!

[Diabolical Artificer]

Apologies.... here we go.

[Radio Wrangler]

Yipes! that thing's throwing away a couple of watts just in the bias chain!

LEDs don't come in handy voltage values. The chemistry of them keeps changing as efficiency keeps getting boosted, so you don't know what current it will give until you start measuring things. They are supposed to be less noisy than zeners, but good practice is to filter the output from a zener.

Whether you use a zener or an LED, you have to bias them with a carefully chosen current. Look in datasheets for a zener's bias current at which its voltage is specified. That's how you're supposed to use it. If the diode current comes from, say 400v then a 5mA specified zener means you'te taking 400*0.005 = 2W but some zeners are spec'd for 20mA and if you don't put close to 20mA down them, their voltage will be low and will not be as well regulated. 20mA would mean 8W waste heat.

Whatever voltage the bottom LED runs at, subtract 0.7v and divide by the resistance of the trimmer and the fixed resistor. That's the current the circuit is set to pull from the valve.

IF the bottom connection on that circuit is grounded and your supplies are all positive ones, and your multimeter negative probe is on ground, then you shouldn't have any negative voltages. Don't just ignore them. They are a sign that something is wrong. Either you have he leads the wrong way round or the negative voltages are real. Negative voltages can be made by self-rectification of strong signals and if you weren't running any intentional signals, that means oscillation. Cheapie multimeters can give silly readings if fed with RF - some expensive ones, too.

Experienced developers tend to use scopes as their primary test instrument rather than multimeters because scopes give more clues if something is wrong, or give more confidence that nothing is wrong. The meter comes out later if I need to check something more accurately, or want a floating measurement across something.

David

[Diabolical Artificer]

Thanks David. I tried to find the current of the zener I'm using without success. It's a "PH C6V8" I did find other similar zeners with graph's but was unable to find I .The CCS is powered - ground at the top or "return" and -78v at the bottom or "input". One thing I'm uncertain about is if a valve has a grounded grid and we run it's cathode from a negative rail, then won't that mean that the OP is then going to be +50v at its positive peak and - 50v on it's negative peak. Won't this cause problems if this stage is connected to another following stage which is powered from 0 -300v say? Won't part the negative 50v (40v) part of the sign get cut off if the next stage has it's cathode at 10v, meaning its grid is at -10v.

I've had real trouble getting a CCS to work as the tail on a LTP, powering it from a negative voltage has worked the best so far, but I'm not really sure why I'm doing this, only that I've seen it done and half understood the reason why. As you say this circuit wastes a lot of power, I think I'll try other types.

Andy.

[wd40addict]

Zeners normally come in various standard power ratings: 400mW, 1.3mW & 5W typically.

If yours is a 1.3W then 1.3W/6.8V = 0.191A.

Have you thought about just using a resistor instead of the CCS? Since you have -75V available it will be a pretty good source!

Incidentally if this is for your 807 amp you could also use appropriate zeners with bypass to allow you to have the HT at 400V yet keep G2 at 300V.