Tuesday, 24 February 2015

A Better HT / B+ Reduction Scheme. Or my problem with Zeners.

 WARNING Please do NOT attempt unless fully competent in working with lethal voltages in a safe and orderly manner! Always PULL THE MAINS supply out of the socket if you want to turn off.

In the UK there are various voltages on the mains, the limits are  ~253V to 217V RMS, though I have seen both limits exceeded. This variance is a pain for valve / tube amps especially when the HT / B+ is already on the high side. Most remedial solutions revolve around  a big Zener diode to the centre tap of the transformer to take a chunk of voltage out.   However a variable mains supply with a constant Zener drop extends the HT variation still further.

A much smarter way would be to put in an opposing A/C into the A/C HT with an additional small mains split bobbin transformer. In addition to dropping the overall HT the line regulation will be better than the Zener option and other windings (6.3V etc.) remain untouched. The transformer I chose is a 250mA 15-0-15 (7.5VA) output from Maplin which whilst cheap and small, is more than enough for the job. The trick is to get the winding the right way round. Using a Mulit-Tap output extends the possibilities.

From measurements, with a 252V mains I have 520V HT. Using the opposing transformer this drops the HT by 55V DC to 465V, much more manageable. If I used the Zener method I could also achieve the 55V drop.  However If I then plug into a 220V mains supply then the Zener will still drop 55V where the opposing transformer would now only drop 48V leading to better overall regulation. Something to watch if you generate the output valve bias from the HT line.

Addition of a 230V - 30V 250mA transformer.

Overall the cost is similar but I get the satisfaction of better regulation.  The transformer should last the life of the amplifier. To locate the correct polarity, wire the opposing transformer to the mains input and connect one end to the low side of the transformer. Then measure the AC between the unconnected output and the HT high side. The voltage will either be higher or lower than the HT depending on the phase being additive or 180 degrees out. When you measure a lower voltage then this is the end to connect to the transformer and the other lead will connect to the rectifier.

Adding the Red in opposition to the Blue results in the Green output.
So where does this leave you if you have a CT (centre tap) full wave arrangement. You cannot just add the winding between the common ground (marked "Point A" below) as the phase means that one cycle will be opposed and the other will be in addition. This configuration requires an opposing transformer with a pair of separate output windings, so a single mains input and say a pair of 0-30 & 0-30 output windings. A 30-0-30 with a single common tap is no good . You then apply the secondary windings as before but in each arm.

A transformer with separate output secondariness for CT Power Supply.

Tuesday, 10 December 2013

HT / B+ Zener Voltage Drop in a Valve / Tube Amp

Note, Updated, please see also  Feb 2015 post.

I am looking to drop the EZ81 out of my amplifier to gain another ECC83 which will result in the HT or B+ voltage taking a hike as the silicon rectifiers drop a lot less voltage in operation than the valve rectifier. The mains transformer I will use has two windings and a centre tap, designed for two diodes to give full wave rectification. The easy and well-trodden path  is to drop ~ 40V of Zener into the centre tap to ground connection of the transformer. It all looks a very simple solution, so how hard can it be?

The choice I have initially taken is to use a string of 6.8V, 5 Watt devices where 150mA of  HT current will equal 1watt of thermal dissipation. The device I have chosen is the 1N5342B from ON Semiconductor.

Below is a schematic of the centre tapped transformer and two silicon rectifier diodes to give full wave rectification.
B+ / HT Zener voltage drop.

Zener choice. Well this has to be more than just dissipation, but that is a good place to start. The 1N5342B  is rated at 5 Watts  but this is at 25C (77 Deg F.). This diode will not be at 25C at the predicted ~1Watt maximum let alone at the rated 5 Watts. The device de-rating means at a 55C (133 Deg F.) body temperature it should only be rated 4 Watts. By using a relatively low voltage drop per diode I can keep the individual device dissipation down.   Leaving the leads long is also a mistake. It is better to get this to a turret tag than leave long wires. Leaving long leads just adds thermal impedance.

There will be a mV/DegC figure and a breakdown voltage vs current attached to the data sheet for whatever device you have, you will need to note these to ensure you stay within your limits. If you have a bias generator from the HT AC rail then this will proportionally track changes in the HT supply and change the bias to the output stage which will  subsequently change the Zener voltage drop as the output stage standing current changes.

Dynamic impedance or what happens when the current through the Zener changes and is also a factor that needs to be looked at. Zener diodes below about 6 volts use a field emission breakdown and at higher voltages avalanche breakdown. Field emission has a negative temperature coefficient and avalanche a positive. The impedance of avalanche is generally the lower but at around the 7V point the diode has a combination of both which gives a dip in the dynamic impedance at this point. So the 6.8V choice is looking good at this point.

Temperature stability is what it is on the data sheet but you can look at compensation if you know what your stability specification is. For example the 1N5342B has a mean temperature coefficient of  +3.6mV/DegC . A silicon rectifier diode (say an 1N4001) has a -2mV/DegC so two 1N4001 diodes forward biased will drop about 1.45V but should leave a residual 0.4mV/DegC temperature drift. Using silicon forward biased in series with the Zener diodes can  reduce the effects of thermal drift. I have already decided that a huge mixed string of diodes is not the solution I will take, but if you go for a single 20 Watt stud mounting device then this is an avenue you may wish to look further at. There are second order effects regarding impedance etc. but by now you can see the principal.

Thursday, 11 April 2013

ECC83 Valve with a low voltage 5V A/C heater (12AX7 Tube)

So why run an ECC83 / 12AX7 with a low heater voltage and what happens when you do?

I have a transformer from an old Linear Diatonic cage amplifier. How old? Well the 32uF+32uF can capacitor has May '58 as the date, but they didn't have just-in-time delivery in those days so I would put it at very early 60's best case. It was originally a two ECC83 + EL84 output with an EZ80 rectifier. It also had an auxiliary 6.3V and HT (B+) socket on the back, so I knew it had some capacity left over. It got put into a guitar amp and became three ECC83 + EL84 output +EZ81 rectifier. (I know, EZ81 is a 6.3V version of EZ80, but I had a new one and it worked OK.)

Now I want FOUR (new preamp + FX loop) ECC83 + EL84, but this may be pushing it a bit, and the last thing I want to do is slowly cook the transformer. If I lose the rectifier valve and go for a semiconductor bridge with a zener to drop the voltage back then I could possibly run the additional valve from the now unloaded 5V heater winding and reduce the total load on the transformer. The light heater loading of ~0.3A should allow the voltage to rise above 5V or safeguard from a low mains supply, but after looking on the net I could find very little in HARD figures. Most advice I saw was that this possibly won't work, with one reliable source saying yes it will be OK. (Valve Wizard)

The only way to decide was to go measure the configuration I was going to use. The measurement below used a DC lab supply for the heater. The voltage supply I have is limited to 210V so I couldn't test with a higher value. The valve was a JJ-Tesla ECC83S with about 50 hours on the clock. I would be looking at the 1-2-3 connected part of the triode. The other half was not connected and the heater was between pins 4 and 9 only. In test this would mean that the valve would be colder than if both heaters were working, but I think of it as margin.

Test Circuit. R3 is selectable as 120k or 240k
Firstly I did some DC testing to bottom out the heater. The heater cold resistance excluding wiring but including the socket was 6.156 Ohm. With a 6.3V supply and ignoring wiring this would draw 1.02A at switch on. In reality this would possibly be lower, but a good (6 Watt) kick to the guts none the less when turned on. When working at 6.3V the single heater draws 155mA equating to a hot resistance of 40.6 Ohm and a dissipation of 0.98 Watt. Working at 5V it drew 136mA  giving an effective hot resistance of 36.76 Ohm with a dissipation of 0.68 Watt which is 70% of the dissipation at 6.3V

With a 1V pk-pk test signal applied to the test circuit I got an output swing of 54.1V pk-pk with an anode resistance (R3) of 120k. With R3 = 240k the swing was 60.2V which is pretty much on the money.  Spice simulation came out as 55.16V  for 120K and 65.5V for 240K, so close enough.

Putting a 4V pk-pk input swing on the input it would drive the output into saturation. This should highlight any obvious weakness in using 5V. Firstly I looked at the negative side as this represents the largest current through the valve.

6.3V vs 5V. Input = 4V pk-pk input 210V supply voltage.

 There is no huge difference to be seen between either heater voltage, the 5V heater gain (blue) looks just shy. The top of the waveform should be well saturated so we will check that next.

As above....

The 5V heater is again a little shy of the 6.3V setting, but the behaviour entering and exiting saturation is pretty much the same. When lowering the input swing and operating in the available linear area the 5V gain is slightly lower but otherwise perfectly useable. Even at a 4V heater voltage the valve is still working in a healthy manner. With a 3V heater voltage the gain has dropped, but this is just experimental data.

So what conclusions does this present. For me and the new guitar amplifier, I'll happily put the first valve with both heaters on the 5V supply. Both stages are below maximum gain anyway. Some voices on the net did say I'll ruin the valve this way, but I found a table in an old design book which showed extended life extending to thousands of hours so I am happy with that.  I will also consider putting a series resistor in all the 6.3V supplied preamp valves from pin 9 to the centre pin of the base and then to the heater wiring. A 2.2 Ohm resistor should reduce the switch on current in the heater from 1 Amp to ~600mA whilst dropping the working voltage to ~5.68V (0.615V drop or ~82% of full heater dissipation) This should have virtually no effect. If it does, then 1 Ohm should be fine.

On the other side of the coin. I should possibly see if I can get a more representative voltage for the supply, 210V is sort of low.


Below is with a 270V power supply with 6.3V and 5V heaters compared. This last test had both heaters wired in circuit. No change from previously observed measurements including at full saturation.

240K and 1.2V input. 73V pk-pk swing.


I have an _OLD_ Mullard Ecc83/12AX7, silvering gone spotty really is shot. God knows how many hours. It still has some output but very non-linear. It normally sits on my desk as an ornament. In the 270V HT / B+ test it performed equaly badly with 6.3V and 5V heater voltages. The 6.3V specification seems to have a lot of margin built in. Just remember to include mains supply tollerance etc. when looking at your own transformer!

Friday, 5 April 2013

DIY Front Panel in Acrylic / Plexi

New amp needs a new face. Looking around in the UK I can get a custom front panel made, but it seems a lot more costly than you can get say in the U.S. Some want .DXF as the file and not having Autocad or a package with this format blocks that route. What I could source very cheaply was some cut to size 1.5mm thick clear acrylic and a couple of sheets of water slide decal paper for a laser copier.

The acrylic was delivered promptly as four sheets cut to size and a set of panel graphics produced using Paint Shop Pro 9. The images were then all mirrored as the plan is to place them on the back of the acrylic and view them form the other side, ie the front. I can then paint over the transfer effectivly sealing them in. The paint observed from the front would have a perfect  finish no matter how lumpy it was put on, or that's the idea anyway. A few old CD cases proved that using finger daubed water acrylic paint.

The laser copier paper arrived with a warning about possibly melting onto the fuser roller of the copier which was not in the advert selling the paper (ebay). However the bigger the copier the more ECO friendly it is and the large business types therefore have a lower than normal working temperature. There was no issue with the printing and it gave razor sharp results. Unlike inkjet water decal paper it needs no lacquer overcoat which is why I chose it this time.

To make the panel it was first taped in place to the front of the chassis. Using an indelible marker the position of the controls was marked onto the back of the acrylic using the holes in the chassis as a guide.  On an earlier attempt I used a scribe which leaves a finer line, but the unevenness made ripples in the transfer on application.

Finished Item...Chuffed to bits!

Test acrylic taped in place and marked.

The transfers we cut out close and then placed in position. The best method I found was to use a white paper sheet under the acrylic with straight lines printed / drawn on. This way you have a datum for the bottom edge of the acrylic and the lines assist in ensuring that the top of the transfer where the wording is is straight.  It took a while to get the result right but here is what I found best.

Lined paper assists getting it straight.

Soak the transfer for thirty seconds and then place to one side for another. Line the acrylic up on the guide paper and place a bit of water where you want the transfer to go. Without this dab of water the positioning time is about twenty seconds. what can then happen is that part of the transfer can grab and when you attempt to move it the transfer will stretch or distort. The bit of water underneath gives you minutes of working time, however at this point it is prone to handling errors if you try to place another transfer as it is effectivly floating.

Using a ruler place as close as possible, note printed mirrored!

After positioning the transfer I would then flip the panel over. The transfer is now against the guide paper. It has enough friction with the guide paper that you can still finely position it if needed by moving the panel. When it is all correctly positioned simply apply pressure, the excess water will be pressed out and the transfer will adhere to the front enough to resist handling. The transfer can still be lifted with a knife edge if required. Having lined paper to do this part also helps but missing in the shot below.

Fliping over to check position and then press to affix.

With the transfers in place I painted the rear of the panel with several coats of artists acrylic paint. I could easily use a spray paint, but the acrylic allows me to test and wipe off. After several coats I did a final coat painted some paper and placed this as a protective backing.

Final coat and backing paper will form a protective seal.

The final result, well really please with this. The edges of the acrylic had some small chipping under the green protective film, but a quick sanding will remove these and they will sit behind the cabinet woodwork anyway.

With painting the transfer blends perfectly to the panel.

It worked out cheaper than having a custom panel, and I still have acrylic and transfers for the rear and a spare front.  One other point about the acrylic paint. It is not fully opaque to light. If I cut a small hole in the backing paper but not through the paint I could shine light  from behind which you would see as a bright yellow glow from the front.. I may not go for a panel mounted power light, but insted use a 3mm white LED mounted behing the panel to illuminate a small circle of blank panel adjacent to the power switch to indicate that the amp is on.

As a final note, If you are thinking if edge illuminating the panel then the transfer edge will become visible.

Wednesday, 3 April 2013

Resistors, Carbon Comp vs Metal Film in Valve / Tube Amp Design.

As in previous post, I don't suffer snake oil etc in circuit design. _However_ whilst shopping for some parts I noticed I could also buy brown carbon composite resistors with nostalgic 3+1 coloured rings for 30 pence. Now they have lots of  know issues but I remembered reading years ago about voltage coefficient in my copy of Horowitz and Hill Sec.Ed.  The question I am looking at is could the voltage coefficient alter the sound of the amplifier? According to H/Hill above 250V the resistor changes value and at 1kV it is out by 29% (Chapter 6, pages 372/3). We are dealing with lower voltages, but across the anode resistor at clipping it could see ~250V potential, so I would like to check that area.

I happened to find an old carbon comp resistor, 680k value hidden away. With carbon composition this  is more of a pre-soldered guide than an absolute value. It was more like 688k.  I decided to do some DC tests to see how much, if any, it would change over a 210V range against a metal film of the same value.

Percentage change with applied voltage.

In the above plot there is a clear change in apparent resistance of the carbon resistor against an applied DC voltage. The measurement contradicts a H/Hill statement that it is really only above 250V where the voltage coefficient becomes apparent. Maybe they meant significant. To put this in prospective, of the original 688k value it represents a drop in resistance of over 20k. The metal film (red plot) is stable at 100.0%.

Having repeated the test and confirmed the behaviour I needed to know if this was something that takes seconds to change or is this change instantaneous and works in real time.

To check this I constructed a potential divider out of two resistors. The bottom resistor was a 62K metal film and the measurement was made across this device. The top resistor was the device under test being either the 688k carbon composition or a 680K metal film resistor. Using a linear ramp source to 210V I could measure the lower reference resistor and determine the behaviour of the top resistor. The tail off at high voltage is the probe capacitance, so linear to +170V.

Real Time Voltage vs Resistance change.

In the above plot the carbon (red) is plotted against the metal film (blue) with the left hand scale. You can see that the red line drops slightly more voltage at the 0.0001s (0.1ms) point as it has a slightly higher resistance value. However as the voltage rises it crosses the blue plot and ends up with a lower voltage drop. This is real time compression well within the audio range.  To clarify the difference in voltage drop is plotted in green on the right hand scale, -7V to +3V. This shows that at the 0.1ms point the carbon resistor is dropping about 1.3V more than the metal film as to be expected being a higher value resistor. However with ~180V applied at 0.48ms the carbon film resistor is now dropping 4.8V LESS. It has shrunk in real time.

Piecewise Linear and Poly Curve Fit.

In a final look at the data you can either view the drop in resistance as a three section piecewise linear (blue, orange & purple lines) with three gradients or curve fit to the equation shown in the title.

The real world effect for valve / tube amplifier designers is that the carbon composition resistor can add some non linearity and compression not normally present in metal film. In my opinion it would be a mistake to consider the use of carbon in the early stages of a cascaded design. When taking the first measurement at DC the digital display would flick around on the carbon resistor but was rock steady with the same value metal film. The noise is really something else. However as an anode resistor in the overdriven /clipping stage, well it may be worth the 30p to someone. I however will probably stick to metal film based on reliability. Is the above effect audible? I will have to say-


Tuesday, 18 December 2012

Technofret Under String Fret Level.

I bought a "Technofret Advance Fret Levelling System" (or TAFLS) from their UK ebay site. It costs £37 including postage (as of today $60 USD). As you can imagine the title "TAFLS" is more than the sum of parts, or sum (cash) of materials. What I got for my money is a single length of  1.6mm aluminium U section channel 480mm long. A straight edge, 5mm thick and about 450mm long. Also included were 3 blocks of Aluminium bar and two grades of abrasive on double sided tape. So far it doesn't sound like I got a good deal for the cash. No instructions or guide were included, but the idea of under string fret dressing seemed whole enough to splash the cash. The real cost is possibly the flatness of the edge and each side of the channel, which are flat against each other and against an optical grade flat at work. It's flat.

Technofret Advanced Fret Levelling System & My String Lifter.

 The idea is simple and an appealing one. The fretboard under tension (allegedly as I don't have information proving this) doesn't bow in a pure curve. With the stings on and the neck under truss rod tension the neck _can_ form a slight S type bend. Using the TAFLS you can level the frets with the board flat under tension which will also (possibly) include any unwanted bows.  The truss rod is then released slightly to set the correct relief and you should end up with a better relief curve. Or so it says on the tin, or didn't in this case.

First part of the process is to get the marker pen out and blacken the tops of all the frets. Usually I do this after I have taped up the fretboard but ideally the blocks used n the levelling process need to rest on the wood so you have to do this with the board unprotected. I managed to get two indelible marks on the fretboard. Doh!

With the strings set to normal tuning flatten the fretboard by tensioning the truss-rod. To check straightness you use the three supplied blocks, one at the first fret, one at about the 17th fret and then somewhere in the middle should do. I  used a small LED torch to gauge the gap. By slowly adding tension on the truss rod you can get to the point that you have no light between the straight edge and the middle block. By moving the straight edge left to right you can also check at what point it can begin to drag on the middle block. One issue is that the straight edge is about 38mm tall and is balanced on its 5mm width on the two outermost blocks. Moving the truss rod will at some point cause the edge to fall and clatter off onto the fretboard if you are not careful. It complicates the set-up but not much.I did check the tuning after moving the truss rod but in my case it hardly changed.

Checking relief at middle block.
Next get the U channel out. It came with two grits of 400 and 600. As I am not in a rush I used only the 600 grit and then used my own tape to place a 1000 grit strip on the other side of the channel. The channel being 1.6mm thick didn't readily fit under the strings at the first fret frets, so I had to lift the strings. The advice from  TAFLS is to use a spacer to lift the strings up, checking correct tune. However I had a small L-shaped bit of alloy handy which I cut to run under two strings at a time. With some protective tape underneath I could easily lift two strings a fraction.

DIY String Lifter

With the frets marked up I did a gentle pass with the 1000 grit. I had already done a traditional fret dress with a 350mm plane as a flat with the strings off, so I was interested to see if the TAFLS idea held up. As you can see below frets 1-4 are being abraded whilst fret 5 (visible) 6, 7 and 8 have a dip and the black marker is still there. Strat necks are more likely to show this behaviour apparently though I have no real solid information myself. This didn't surprise me as the guitar did fret out on the first few frets if the action was set too low.

Fret five, still black

 I would step across the neck one string at a time from low E to high E doing a small circular motion each time and then turn the U channel round and repeat the process from high E to low E. I would always lift the strings in pairs to make room for the U channel flange. After getting down to all frets showing abrasion I turned the channel over and used the 1000 grit to complete the initial dress.

After completing the dress the fret ends were re filed and the crowns re profiled. The frets were polished to 2500 grit and then a Dremel and some metal polish to finish off.

 At the end of the process I can say  that the TAFLS worked a treat on this neck. This said, a 7.5inch neck will never be as low as my other guitars with a more forgiving  radius I can say that this time it was better than before and the action is now as low as I can hope for allowing bends not to fret out. Possible I didn't flatten the fretboard in my previous attempt but despite feeling it was expensive for what I got, the result made it money well spent.I do worry that the channel doesn't feel robust so I have a separate place to store it. Just for the record, I don't have any association with the source of the TAFLS other than I purchased one for full price. There are other questions I would like answered. How much does adding tension to the truss rod change the neck, should it be a combination of added tension and slightly de-tuning the strings? Where is it best to place the middle block and in the case of large block inlays, is it OK to use fret 8 where there is no inlay? It isn't the only under-fret, under tension method available, but it was local and not too expensive. Maybe it's a fashion that will soon disappear, but I can see the logic of dressing frets with the neck under tension.

Friday, 21 September 2012

A New Valve / Tube Tilt Tone Control.

This is very much an experiment at the moment. Years ago QUAD brought to a wide audience a very handy tone control for their HiFi preamps It was termed a Tilt control but is also known as a contour control which I find less descriptive. It differs in being a shelved filter so you control a broad range of frequency by a bit rather then a small area by a lot. It is better at changing the emphasis of the tone and not the tonal character. The idea is appealing but it is normally wrapped around an op-amp. I could passively add it but I came up with a topology which was far more tempting.

The topology places the control within the phase splitter and the output stage in my EL84 amplifier. The output is fed back into one side of the phase splitter. I liked the idea as the circuit has a nice symmetry which appeals.  The complication is it sits within the feedback loop from the speaker to the phase splitter so it's was always going to be a bit of a battle to ensure stability. If you look at the schematic below you can see that it would be very easy to add to an existing amplifier as it sits on top rather than within a normal EL84 output stage. The other issue I have found is that the switching from Pentode to Triode it changes the feedback gain and hence gain of the tone control.Click on the schematic to see it full size.

The frequency plot below is a multi-run simulation where the  TILT control is moved from 0k to 470k in linear steps. It shows clearly how at one end of travel it has bass boost and treble cut and at the other bass cut and treble boost.

In the circuit R47 (33K) sets the amplitude of the boost and cut in conjunction with R48 the feedback resistor. Current work is refining the balance between the two values which results in a good degree of tone control without sacrificing stability. It has proven to work well as a tone control but it would be better if the existing bass-mid-treble was not directly prior to this stage in my test amp. Better would be a topology closer to a trainwreck circuit where the tone control is early in the preamp cascade. That way you could alter the tonal emphasis going through the distortion stage and correct in the power stage.

If you are content with a boost / cut of a few db then phase shift can be kept very low, something that HiFi designers may find useful, which brings this full circle. It is trivial to obtain flat response at the bass end to low Hz.

EDIT. Looking at the above simulation the overall gain is ~29dB with flat gain, it should be nearer 18dB. I guess I took my eye off the gain whilst optomising tilt. The omission was to leave the grid capacitor C14 not tied into the feedback point. I'll address this excess gain and see if the instability issues previously noted disappear.