Handbook Amp Project – Part 3 – Rectification

Print Friendly, PDF & Email

The Rectifier Diodes
For valve or most transistor and Integrated circuit based appliances like radios and PCs to work, high 240V mains voltage usually needs to be converted to DC that has no varying aspect (ideally) to it. This is done by components in the Rectifier section of the circuit. The rectifier section converts a varying AC sine wave to an (ideally) unvarying DC voltage so that current flows in only one direction through the circuit – from positive to negative – i.e. Conventional Current flow. This is a convention for historical reasons, before the electron was discovered – actual electrons flow from a negative source cathode to the positive anode.
A diode is a component that only allows electricity to flow in one direction but not the other.
        
When one is used in an AC circuit, the current can only flow in one direction for half of the AC cycle at a time – so it acts as a converter of AC to varying voltage DC – this is half wave rectification, and is only 50% efficient as only half of the wave is flowing in the circuit for half the time.
http://en.wikipedia.org/wiki/Rectifier#Half-wave_rectification

There are other types of rectification offering more efficient full wave conversion methods – using 2 diodes (requires a centre-tapped secondary HT winding) or 4 diodes (aka Graetz Bridge).


There are two main types of basic rectifiers for amp projects – a valve based design of diode plates housed in a glass vacuum tube like the GZ34:
        
or discrete PN junction component diodes made from silicon or germanium. These diodes can be single or many, combined in a sealed housing.

The 5EY equivalent – the GZ34 – I have, but will not be using in this project as it requires a centre-tapped secondary – is the:
http://www.ampmaker.com/store/JJ-GZ34-rectifier-valve.html
http://www.r-type.org/exhib/aaa0013.htm


Probably the most common types used are the 1N4001 – 1N4007 range of diodes rated at up to 1A and 1000V Repeat Reverse Voltage:

Part of the datasheet:

So for valve amps, the 4007 is a good choice, as used in the Aston kit, though any from 4005 up would do, as the Aston has no voltages above 400V in the circuit. A+ was 335V.
Ok, after a component check for this section before wiring it, I realised I can’t use the GZ34 rectifier for this project as the Danbury transformer is NOT centre tapped – Duh!. I didn’t read the book parts overview properly before ordering the components list, as I also now realise the Handbook Project recommended transformer (Weber W022772) has a separate 5V, 4A winding just for the GZ34 rectifier diode heaters – the other 3 valves use the usual 6.3V winding. The Danbury does not have this extra winding. Well, I did say it may be a bit of a hybrid! Do your component research before you buy I guess! Don’t assume all valves are similar much less the SAME!
Also, I only seem to have 2 ceramic Octal valve bases – not 3 – I have an extra Nonal base instead – so the rectifier will be dropped now anyway, as I want the option of running 2 x 6V6 valves in parallel or 1 6L6 – though I have found these amp kit components are pretty flexible to voltages anyway as I have run a 6L6 and an old 1956 radio RCA 6P6, higher voltage valve in the Aston and they worked fine. The 2 x 6V6s should give more power out so it is the current rating at full volume that is important if running 2 x 6V6s. The secondary is capable of about 160mA max according to
http://www.ampmaker.com/store/0-190-275V-20W-power-transformer.html
I will have to order more mains rectifier diodes, and an extra Nonal base to cover the last of the 5 chassis holes also.
I also read today, that valve rectifier “sag” which I know little about (so wanted to learn by using a valve diode in the first place) may be a non entity today due to mains line voltage. From this forum:
http://music-electronics-forum.com/t26515/
This comment got me thinking what it involves:
“I’ll take your word that true GZ34s may have had less voltage drop, but (at least in North America) line voltages are also considerably higher than back in the day, so it may be essentially a wash as far as A+. A small difference in sag is probably all you might have comparing Old-days GZ34 and modern 5AR4.
In any case, the voltage drop with either is small enough that there is a very marginal argument for using a GZ34 class tube to begin with….there’s so little sag you might as well just go solid state for the recto.”
If it was decades old lower and less consistent mains voltages that caused this “desirable” sag sound effect in tube amps – a form of compression – (due to the power section drawing all the available output current across itself at the expense of the rectifier section I assume?), then it probably doesn’t really apply today, and it would only be noticeable at high output volumes anyway? In a well designed amp I guess sag should not happen in the first place as all components should be able to supply what is required of them. As stated above, it seems it was the mains line voltage drops that caused the effect – not an intrinsic “design fault” with the older amps.
I guess I’ll have to wait for a 3rd Amp Project build to use the GZ34 and find out…
For now, this Project has taken a turn toward the Weber Maggie Amp, which is heavily drawn from the Handbook 2 Stroke Amp anyway apparently, now I have found a circuit diagram for it:

This is how the 4 diodes make DC in the same direction across a load at each alternate ½ cycle:

NOTE – B+ error above – it should be A+. It is the highest DC voltage point in the circuit once the first and largest smoothing cap is added in parallel to the load.
Filter capacitor and load resistor values
The 220k load above also represents the bleed resistor or rest of the amplifier logically – all the amp DC components will sit between A+ and Earth.
Before showing pics from the oscilloscope, I better explain what is required to measure mains 240V AC voltages, and up to 400V peak DC.
As the voltage/division knob on my scope goes to 20V max, (though the actual scope input connections can handle 400V DC or 400V max AC + DC mixed), you need a probe that attenuates the input signal so it fits on screen on the 20V range. 10 and 100 times attenuator probes are available. You can check the probes accuracy with the 0.2V and 2V calibration signal points on the scope, with the settings on 0.2V/div and 2V/div. The square wave pulses should sit bang on the 1cm line for each setting. If not, your scope may need calibrating, or your probe is a little inaccurate.

The important thing to know when measuring mains level voltages is to use ONLY the probe tip, and leave the black croc clip disconnected if measuring across transformers or mains. If you don’t, you get a peak to peak voltage RMS value of +/- 240V = 480V peak to peak value to the scope input – say from connecting across the mains live and neutral – and this exceeds the input rating of the (my!) scope (depending on your model etc. of course). This causes the scope overload protection to short to Earth and trip the RCD (hopefully) without any damage.
You may ask how you get a reading if you don’t complete the scope input circuit with both red and black connections attached. I wondered this also, until writing this now, and I THINK the relative point is chassis Earth anyway (so you can connect the black clip to this) – because the scope’s own chassis is also earthed, so your live probe point will always have a reference point – even if the probe is touching a different bit of kit.
If you don’t use a 10X probe on a 240V line, the signal is way too big to read on the screen. It is still within the input max of the (my!) scope though.
Post Diodes, Pre capacitor connection DC readings.

Above – the measured Multimeter DC from the 190V secondary after full wave rectification by the diodes only. Although the cap is on the board, it is not in circuit at this point as the Standby switch is OFF.
Below – using a 10 x PROBE for the 190V secondary diodes reading

Above – the unfiltered full wave DC from the diodes only. 10V/div x 10 Probe = 0-200V varying DC. Note the kinks in alternate ½ waves – is it due to the greater non linearity of one diode pair conducting for their ½ cycle? It looks like one diode only that conducts a bit differently to the others.
OK, before I go any further, you should read the Earth Grounding PDF from the Valve Wizard’s site:
http://www.valvewizard.co.uk/Grounding.pdf
It is a good point in the Project to have discovered this PDF, as it explains really well, some important design features an amp build should follow from an electrical theory point of view.
The chassis Earth points should follow the general layout of the components for the various sections, to keep noise to a minimum – all else being equal. The basic logic is you would not have your guitar signal jack socket connection running next to the mains transformer, near its magnetic flux field before it gets amplified, on purpose would you? You place all components logically in this sense also where possible. Read it yourself anyway…if you get mains hum after your build, this may be the first place you look to solve it.
The 220k Bleed Resistor
Using the Aston as an example, R19 is the first voltage drop resistor, that would pass only about 220V/220K = 0.001A or 1mA.
This resistor R19 allows the high voltage at C13 to discharge completely to Earth eventually, once the Standby or power switch is turned off.

An important thing I have discovered so far in my very amateur and limited personal experience, is that all the amps I have worked on so far actually drain of their high DC voltages quite quickly once switched off – IF WORKING CORRECTLY. I have used my multimeter on the 1000V DC range to check the DC at the first smoothing capacitors of the Marshall, Aston, Ashdown and Fender amps and found they are almost at 0V within minutes of power off at worst. The Aston takes 2 mins to get below 10V.
The time constant for an RC circuit is the time it takes a cap to discharge to about 63% of the value difference between the initial value and final value. It is an exponential decay:
http://en.wikipedia.org/wiki/RC_time_constant
T = RC
For a 47microF capacitor and a 220k resistor this gives T as:
0.000 047 x 220, 000 secs = 10.4 secs
This means the cap should drain very quickly from disproportionately higher values, getting progressively slower at lower voltages. Beware bigger capacitors and with bigger resistors!! They may store larger charges for a lot longer.
With the Standby switch connecting the 220k and 47microF cap, I measured 281V on the meter (the big 5W, 10k resistor you see is not connected to 0V so effectively out of circuit – just a convenient Probe point):

I plotted this period very roughly for these components in Excel, at about 5 sec intervals for these values from 281V DC at A+, while watching the DMM reading drop and got:

This is reassuring that these components drain to less than 10V DC within 50 secs of power off. Even after only 10 secs, they are below 100V. Within the first 5 seconds it drops by 150V.
HOWEVER, this does not mean that you don’t ALWAYS check for dangerous voltages before manually working on any amp or electrical device!
It also does not mean ALL other amps are like these in design, or don’t have a fault which stops the capacitors discharging through a bleed resistor – e.g. a resistor faults open circuit – like a resistor did in the Marshall, but at the HV pre-amp stage.
Another bonus I realised, is using the AC neon bulbs across the Standby here also, as they stay lit until most of the voltage has drained away, slowly fading out visibly to nothing, at about 80V DC.
An important point here that may confuse beginners to electronics – it did me (still does!) – when I look at a circuit with DC blocking caps, to know what voltage is there.
Once the cap has charged up, no more DC current flows, so there is no motive force pushing current through any chained resistors above it, so no theoretical voltage drop between them. In practise, there is a difference, as I found after testing. I don’t know what causes this at this point, so it is an Unresolved Question that I will be logging at the end of each section Post.
The + plate of C9 now becomes the same value of the prior resistors value, once all caps have charged up= A+ line voltage, less small practical differences. This is why I could connect the probe to the end of the disconnected 10K resistor in the pic above and read the same value as A+ there. The same should happen (nearly) when the cap C8 connects the 10k resistor to earth, once it has charged up.
These voltage phenomena become apparent at the end of this section’s summary readings at A+, B+ and C+ points.

With the Danbury at 300V max secondary, let’s do some theory as to why a 220k, 1W resistor can be used here, or as a safety test DC drain lead like the one I made up for my first build project (below), from doing my research before getting too involved with valve amps.

Two very common equations used in basic electronics is Ohms Law, V=IR, and a Power equation:
P=IV
or re-arranged, as V=IR already from Ohms Law,
P also = I x I x R
or, as I = V/R,
P also = V x V / R
So, as the Danbury secondary is nearly 300V RMS AC, this peaks at about 300V x root 2 = 300V x 1.414 = 424V. This is the peak voltage the first rectifier capacitor will try to charge to, more or less, before it drains a little, as the voltage cycle drops between charging peaks.
From Ohm’s Law, the peak current through the 220K resistor is therefore 424V / 220, 000 ohms = 0.002 A or so.
From P=IV, the power dissipated through it is about VI = 2mA x 424V = 0.848 Watts, so a 1W rated resistor is sufficient. As most small valve amps are usually below 500V DC (the Fender Champ 12 has the highest A+ so far of my amps I have measured, at 505V DC) a drain lead with a 220k, 1W resistor will still do. 500V/220000 ohms = 2mA. 2mA X 500V = 1.1W. Within +10%…
In reality, I just use the Multimeter on the 1000V DC range – it may be slower draining due to the meter using a higher value internal resistor chain, but at least I can SEE the actual voltage dropping to a safe level. Who is in a rush…?
As a comparison, here is the 60W Fallen Angel rectifier section, which also uses a 220k resistor R87, after the 4 rectifier diodes, across a 270V RMS secondary like the Danbury, but would have a much higher secondary current rating to feed 60W output, and all the other features and 2 channels of this amp:

This circuit also has interesting mod options of the extra capacitors 62, 63, 65, and 66, that I assume are for higher frequency anti-ripple as they are a very low 4.7nF value?
I will have to research that circuit addition. Hmm, no info so far – maybe just try it and scope the results after the main build…? (Unresolved Question)
Anyway…
Use of a single 1N4007 for ½ wave rectification, is as the Marshall circuit’s D1 – diodes must have been expensive in 1972!

Also, note the primary fuse rating of 500mA – halfway between the Aston and the PP18 amps. Note that the primary fuse rating doubles for a halving of mains voltage – for the US or Europe etc. The Aston label states 250mA fuse for 240V UK mains, and 500mA for 120V mains.
Again, VIprimary power used in both cases is the same; 240V x ¼A = 60W, or 120V x ½A = 60W being used at the fuse limit in both cases.
Hmm, that’s an interesting thought too then… If the primary uses a maximum 60W in the hypothetical 100% efficient transformer, then 60W is used in the secondary also, but if the speaker power output is only 20W max, then 40W is being used just in the components themselves – a lot of heat lost for little sound out! That’s quite inefficient eh? Or have I overlooked something fundamental, as I do sometimes…
….Er, yeah – I think the 60W is total amp power, so 30W each for primary and secondary sides, so only 10W, not 40W is being “wasted” in the components for a 20W speaker…?
The Marshall does not have an equivalent of the single 220k to Earth as it is a slightly different design, with DC blocking/filtering capacitors from X (C17, C18 and C25) to Earth.
These first 3 caps can be re-drawn for clarity, so it is similar to the Grounding PDF filtering example with no 220k:

This way you can still follow which point feeds the pre amp/tremolo and power amp sections
From a discharge perspective, it does have 3 x 13.8M ohm or so voltage divider chains in parallel across A, such as the chain made up of R5, 3 and 4 below. You can follow the other 2.
Note that this still allows point X to drain all HV capacitors C17, C18 and C25 eventually when switched off also.

http://www.learn-about-electronics.com/capacitor-filters.html
http://www.circuitstoday.com/shunt-capacitor-filter
http://www.docircuits.com/circuit-editor
The Weber Maggie amp circuit that is very similar to the Handbook kit – nice and simple – the next amp build, (after this one), may be almost a clone of this Maggie, depending. The present amp is a combo prototype of both and a learning platform, and to use the component values I already purchased for it:

Rectification Pics
So, summary so far – I have rectification at A+ on the scope pic – still VERY bumpy DC with 0-200V peaks in it. No Standy switch on, so no cap or 220k in circuit:

Note that if you use the scope DC switch, it causes a drop of 20V on the DDM, from 207V DC to 187V DC. Here it is on the AC setting to show just the AC ripple component of 200V p-p, and so the DDM DC reading gives the 208V DC. 10V/div x 10 probe settings.

What happens with the addition of the 220k and 47microF cap at A+? The measured DC using the 190V secondary was 281V:

The ripple now is this. 500mV/div scale x 10 Probe = 7500mV or 7.5 Volt or so:

The first capacitor has smoothed out nearly 200V of ripple to just 7.5V!
So, time to add the 10k, 5W resistor and the 16microF cap and see what improvement to the ripple this makes:

It doesn’t change ripple at A, but it causes a drop by 2-3V DC. The voltage at B+ (C8) is now 270V, a drop of 10V or so from A+, and the ripple is 0V.
Adding the final 68k resistor (the value for the Handbook circuit, not the Weber = 56k) and 10microF, 250V rated C9 (closest value to 8microF) I get these voltages for
A+ 277V
B+ 276V
C+ 275V
I now get about 750mV ripple at A+; 50mV/div X 10 Probe:

Tiny 25mV ripple at B+ and C+, 50mV/div x 1 Probe (Chan 2):

The Aston ripple at A+ is about 400mV, so about half.
The main (future) issue is to change the 10microF, 250V cap as the voltage here is 275V, so it’s over its rated limit, but working.
If I wish to experiment in future with switching in the 275V secondary winding, this cap will blow with an extra 100V or so more across it. Ampmaker.com doesn’t sell 8 or 10 microF electrolytic at 450V ratings, neither does Maplins.
I may swap it out for another 16microF, 450V instead. This won’t not make much difference overall.
So why these particular cap values in the first place? (Unresolved Question)
As the idea of DC blocking caps is to hold the secondary DC voltage supply line as close to maximum as possible for the three sections A, B and C, they are designed to power, why have different values at all? Also, if a general rule of thumb I have read on forums and elsewhere is use large caps for a greater reservoir of charge (so better smoothing between cycles), and they pass lower frequency AC, like mains hum also – say 100microF caps – why not just use 3 of these instead?
The only thing I can think of, in my limited knowledge, is these values offer a compromise for these two mentioned desirable functions, whilst passing some higher frequency noise also – maybe 100Hz half cycle bumps – which the last cap certainly seemed to remove from A+, where the 2nd cap alone didn’t.
Research required.
Unresolved Questions

  1. What are the small 4.7nF caps for in the Fallen Angel rectifier section for?

These are called snub caps, to filter high frequency switching transients as the diodes suddenly turn on. These transients can leak onto the amp chain, possibly from the deformed heater sine wave leaking across to the cathode of the valves. This noise causes a nasty form of AC noise called Rectifer Hash, and is very buzzy, from the spikes it causes at the anodes of the valves. 
        2. So why these particular filter cap values in the first place?
The initial value of say, 47uF is calculated from the max current rating of the PT sec, so as not to cause too much inrush current to flow into the uncharged cap at switch on, and cause the PT sec to overheat or burn out.
http://www.valvewizard.co.uk/smoothing.html

In the following example we have a 50W push-pull amp, and the HT after rectification is 350Vdc. We also need to know the average current the amp will draw. This will be the sum of the quiescent currents of all the valves.
Supposing this amp has two ECC81’s, an EF86 and two EL34’s, the average current drawn by each triode will be about 5mA and there are four triodes in all, making 20mA. The EF86 will draw about 3mA, making 23mA for the pre-amp.
We will assume that the power valves are biased to their maximum dissipation of 25W each (if the amp is Class-AB they won’t be biased that hot, but we should assume worst case scenario). The average anode current they draw will be in the region of:
I = P / V
I = 50 / 350
= 143mA.
The screen-grids will also draw a quiescent current: The data sheet for the EL34 suggests a screen-to-anode current ratio of 6.5, so we can expect the screen currents to sum to:
143 / 6.5 = 22mA.
Added to the pre-amp current this makes 188mA in total for the whole amplifier. (Remember, this is the averagecurrent, and is not the same as the peak current that the amp will draw. In an amp like this, depending on the class of operation, the peak current might be 200mA for one EL34 (while the other EL34 goes into cut-off) plus the 23mA pre-amp current which won’t vary much. That makes 223mA peak, and the power transformer would need to be rated for at least this much AC current, preferably 1.5x more if the amp is to be run at full power for long periods.
We decide to allow 10% ripple. The HT is 350V, so 10% of this is:
(350 / 100) * 10 = 35Vp-p.
The reservoir capacitor can be found using:
C = (t * I) / V
Where:
I = average load current drawn
V = ripple voltage peak-to-peak
t = duration between charging cycles and is equal to: 1/twice mains frequency.
In Europe the mains frequency is 50Hz, so: t = 1/100 = 0.01 seconds.
The reservoir capacitor required will be:
C = (0.01 * 0.188) / 35
= 54uF.
The nearest standard is 47uF. This would give a 40V ripple which is 11%; perfectly adequate. Most modern designs use a value from 33uF to 220uF at most. Hifi amps may use much more, but don’t be tempted to use very large values unless your power transformer is designed to handle a lot more current than you actually need. The low power-factor caused by using a huge reservoir capacitor can cause a ‘borderline’ transforer to overheat! If using a valve rectifier you MUST check the data sheet to see if the value of capacitor is allowable [see below].

——————————————————————————————————
Misc Build Notes From Web:
http://www.electronics-tutorials.ws/diode/diode_6.html

Research amp Earth Loops:
http://www.valvewizard.co.uk/Grounding.html