Single ended amplier

[Diabolical Artificer]

Yesterday I replaced the OPT with a "choke" tfmr and got a nice clean OP at 3.2W. This choke tfmr has taps on the primary for use as HT smoothing.  One part of me thinks this is a neat idea, but I have a feeling someone will tell me it's not. Anyway, this is academic as I won't be using this facility. The OPT ratio I'm using ( it has sec taps to alter the ratio) is 33:1, so with a 8 ohm load that gives me an 8k ish primary.

Michael, wouldnt a mains tfmr saturate?

Yes your right Jeffrey, we don't need all the digits. One thing I like about valve amps is that we don't need to be too precise, ish will do.

Thanks Kalee, that helps a lot. Re the one meg pot, I often read posts/threads and get the gist but don't remember all the detail. I'm reluctant to to use a bigger ECC82 anode R as this will lower the load line, which will put it through the curved part of the g1 line. AFAIK, this will make the OP non linear and introduce distortion. That's not that important here in this amp, as it is hardly super hifi, but still. I'll try it later and report back. BTW, some mutual characteristics graphs use 8 squares per 100v/10mA or what have you instead of 10, which is a pain.

Mark, I'll read your paper with interest. This "frequency/dB/node" subject is one I'm having trouble with. Sentences like "components act as a high-pass filter with a -6dB/octave (-20dB/decade) slope and a lower -3dB point " leave me scratching my head and looking puzzled. Why is it always 3dB? Octave/decade??? I've tried to get my head round this , Kirchoff and other exotic subjects but it's heavy going. Hence this amp and the use of formulae etc. I'll start a thread about dB/frequency etc.

Now i have an OPT that is ok, I'll re wire the amp today taking into account all your suggestions and advice. I've posted some pics of the load lines I did as Lawrence asked for them.
the EL84 is a bit messy but it should be about right. I've found drawing a load line for and getting my head round some of the aspects of SE OP harder than for P-P.

Thanks for your input, Andy.

[Murphyv310]

Hi.
In the recent overhaul and rebuild of the World designs amplifier It didn't have any form of volume control it relied on the 1 meg grid resistor and 4.7k grid stopper.
During the time I did the work I didn't have the customers preamp.
When he got it back I got a call from him he told me the Bass was brilliant. A few weeks later he came back with the amp, speakers and preamp for an intermittency on the left hand channel (turned out to be his £200 phono leads) but I had a look in the preamp (solid state) and I was surprised to see the output coupling caps were measly 0.047uf. This of course was one of the reasons when the amp had been butchered on the few occasions it worked produced poor Bass.............. a 47k across the inputs. Now don't ask me the make of the preamp all I can say it was well over a £1000 and was designed for use with valve monoblocks.

I accept that a high value volume pot "can" lead to hum pick up especially in designs that may have a far from ideal layout but my point is that in high impedance inputs a 50k would be the wrong choice.
Yesterday I quickly knocked up an amp using the values the OP has used and yes it does work but I found that using line level there is no headroom at all and there was some instability, Raising the 12k to just 47k and increasing the cathode a couple of hundred ohms cured the instability and improved the drive no end, it still is bang on in the middle of the curve and on my distortion meter the overall output at 3 watts was under 10%, later on I reverted to an ECC83 and higher anode load with some NFB and the distortion was less than 5% the values were selected for best linearity.
The HT was 280v in this lash up.

[Refugee]

I am a bit spoiled in a different way as I have a pair of monoblocks that use a single triode phase splitter and make use of the other half of the double triode as an extra gain stage.
For 3.5mm outputs on gadgets a 33 ohm resistor hidden in each cable at the amp end kills all hum pick up from the cables.
Using a DVD player as a CD player has the opposite effect and potential dividers have to be hidden in the Scart connector.
As found the monoblocks just had a 1 meg resistor hidden in the top cap of the preamp pentode. I added a 100nf SMD cap in with the 1 meg resistor to prevent offsets from horrid gadgets from driving the preamp into saturation.
It works fine.

[pwdrive]

Andy wrote "I'm reluctant to to use a bigger ECC82 anode R as this will lower the load line, which will put it through the curved part of the g1 line. AFAIK, this will make the OP non linear and introduce distortion."

So far as I understand these things a lower value of Ra will produce more distortion than a higher value of Ra, are you confusing mutual characteristic curves with anode characteristic curves? (I stand to be corrected)

I've done a couple of load line plots below for the ECC82, one for a 12k anode load and one for a 50k anode load to show the difference in anode voltage swing either side of the operating point, for convenience, an operating point of -4v is chosen, the distortion figures are approx only and are based on a formula in RDH4 (Langford Smith).

You'll notice that the anode voltage swing either side of the operating point also includes a distance measurement in millimeters...! I find it an easier and more a precise method of working out the voltage differences on those types of data sheets.

       

Lawrence.

[Mike Watterson]

Yes, 47K much better than 12K, also probably much longer valve life.

[Kalee20]

DA, I have looked at the load line you have kindly posted for the 12AU7 (ECC82), the 25kΩ load line correctly runs from (0V, 10mA) to (250V, 0mA).

But at your operating point, (138V, 4.5mA) approx, your swing for 25V (total) out is 1V on EACH SIDE of the -5V grid voltage. So gain is 25V / 2V = 12.5, not 25! This ties up with the ECC82 having a µ of 17, which is the maximum gain that can be achieved with RC coupling.

Ref the discussion about more kΩ in the anode load bringing us down to the curved foot of the characteristics, well yes I can see what you mean, but what matters is the distance along the load line that the that the curves intersect with the load line. If these intersectioins are equidistant, then distortion is peanuts. Lawrence pwdrive has illustrated this - his AB is equal to BC pretty nearly.

The anode characteristic (Ia against Va) for constant Vg - which is what the your graphs show - is indeed curved towards the foot. But it turns out that the mutual characteristic (Ia against Vg) - for constant Va - is curved the other way such that the two cancel out. And that's why the points of intersection referred to tend to be equidistant.

The µ of a triode is a pretty reliable and constant characteristic. But mutiual conductance gm isn't. So if you use a really low value of anode load (like 1kΩ) the gain is going to be small and virtually dependent on gm, with the prospect of distortion as you run around the foot of any curve. Whereas if you use a very large value of anode load, like 100kΩ, or even better a constant-current load or a really high-value inductor, then the gain approaches µ which, being constant, implies low distortion.

[Mark Hennessy]

Hi.
In the recent overhaul and rebuild of the World designs amplifier It didn't have any form of volume control it relied on the 1 meg grid resistor and 4.7k grid stopper.
During the time I did the work I didn't have the customers preamp.
When he got it back I got a call from him he told me the Bass was brilliant. A few weeks later he came back with the amp, speakers and preamp for an intermittency on the left hand channel (turned out to be his £200 phono leads) but I had a look in the preamp (solid state) and I was surprised to see the output coupling caps were measly 0.047uf. This of course was one of the reasons when the amp had been butchered on the few occasions it worked produced poor Bass.............. a 47k across the inputs. Now don't ask me the make of the preamp all I can say it was well over a £1000 and was designed for use with valve monoblocks.

I accept that a high value volume pot "can" lead to hum pick up especially in designs that may have a far from ideal layout but my point is that in high impedance inputs a 50k would be the wrong choice.

As I said earlier:

...LF roll-off in this context is academic with modern gear. Different if you're building an amp to work with valve sources (that are often too mean to include a cathode follower and a decently sized output cap)...

So whatever this preamp was, the designer was clearly too mean to include a decently sized output cap

That is simply bad design - it's very wrong to make assumptions about the following equipment, especially in something that was clearly designed this side of 1980. It happens, sure, but it's bad. I wonder what Zout of the preamp was? Hopefully not so high as to cause HF loss in the interconnect...

Incidentally, the Quad FM4 is well-known for its miserly output cap: http://www.audiomisc.co.uk/QuadFM4/FM4.html

Yesterday I quickly knocked up an amp using the values the OP has used and yes it does work but I found that using line level there is no headroom at all and there was some instability, Raising the 12k to just 47k and increasing the cathode a couple of hundred ohms cured the instability and improved the drive no end, it still is bang on in the middle of the curve and on my distortion meter the overall output at 3 watts was under 10%, later on I reverted to an ECC83 and higher anode load with some NFB and the distortion was less than 5% the values were selected for best linearity.
The HT was 280v in this lash up.

This is really interesting - thanks for doing this

By "no headroom at all", do you mean the triode section was clipping before (or at the same time as) the EL84?

What do you mean by instability? Oscillation? What sort of frequency?

The distortion is a bit better than I'd expected.

I think Andy's done a really good job here, and it's obvious he's thought everything through in detail. And now that he's starting to tame this beast, we can discretely mention that the downside with his carefully selected triode stage is there is no scope to use negative feedback to linearise it. The distortion readings from your work certainly show a worthwhile improvement from the use of NFB.

From those that know, any comments about the original transformer? E.g., is it actually intended for push-pull operation?


Lawrence and Kalee20,

Thanks for your posts, which are helping to make things clearer for me. I've always struggled to interpret load-line graphs, partly because I've never see the value of them for transistors (to the extent that I've often wondered if they are a "hangover" from the valve days, where the load-lines are clearly very different, and there is information to be found there).

What I'd really like to see is a graph of gm verses Ia. Do these exist? Perhaps I'm too set in my "transistor ways", where gm is basically 35 times the collector current (simplification, but good enough for all practical purposes IME). And I'm spoilt by bipolar transistors, which have so much more gm compared to anything else

I've made a bit of a boo boo when I said the OP of the ECC82 is 25v, it is 25v but that's P-P, so that would be 8.8v rms.

Worry not - there's no boo boo, Andy. Your diagram clearly says 25V pk-pk

Keep at it!

[Mike Watterson]

Power FETs can usefully use load lines, so actually can bipolar!

Many bipolar transistors have a pentode like Ic vs Vc (instead Ia, vs Va), i.e. a relatively constant current vs supply voltage above the "knee" and many FETs are more like triodes, or power VMOS types more often like voltage controlled resistors (linear output current vs supply voltage).

Loads on valves are usually resistors, transformers or inductors, except for cascode circuits. Transistor "loads" can additionally be totem pole (rare on valves) or complementary driving resistive load with centre tap supply or simulated via output coupling capacitor. A load line is most meaningful for Class A, which is very rare on transistor o/p stages, as they are usually push pull (yes transformers on pocket radios to amps for 100V line PA), totem pole/quasi complementary or complementary class B or AB. Valve push pull seems to be more often AB (usually transformer), though I've seen SOME class A and class B.

Yes, you can plot gm vs Ia (or anything else) on a spreadsheet from the data from uTracer. It allows you to save text files of all the data captured, though the built in graphs are pretty clever. It may even plot gm vs Ia directly, I don't remember. It can display user defined load line.

[pwdrive]

Mark wrote "What I'd really like to see is a graph of gm verses Ia"

1st graph:

http://www.r-type.org/pdfs/el84.pdf

Just a general note: To determine mutual conductance from a set of anode characteristics.....Draw a vertical line at the voltage of interest so that it intersects two grid voltage curves, that represents a change in grid voltage for a constant anode voltage, draw a line from each intercept point across to the anode current, that represents the change in anode current for the change in grid voltage.

Subtract the lower grid voltage from the higher grid voltage, subtract the lower anode current from the higher anode current, divide the resultant change in anode current (in milliamps) by the resultant change in grid voltage to obtain mA/V.

Lawrence.

[radioman]

From those that know, any comments about the original transformer? E.g., is it actually intended for push-pull operation?

The RS output transformer that the OP used (217-567) should have been ok - however, I think it would be a good idea to measure it's primary inductance with a meter. I also have one of these 217-567 transformers and it's your standard gapped lamination type for single-ended use although it's rather small physically and is only rated at 3.5 W.
The one currently on sale (RS 210-6475 made by OEP) although it's got the same published specification is not made in the same way...
I ordered a number of these a few years ago for making up some simple S-E amplifiers and was very surprised to discover that it had interleaved E & I laminations which would imply it's for push-pull operation. However the data which is shown above and which is still available to download from the RS website still mentions the 40mA primary current max. and 3.5W max. output also implying it's for S-E use with no mention of push-pull.
I was so concerned about this when I first got them that I contacted RS technical helpline but when I raised my concerns to their 'expert' he had no idea what I was talking about....
I mentioned this problem with this transformer on UKVRR forum at the time but was told it didn't matter but I think it does.

Andy