New valve amp build

[Solidus]

It's me again, the know-nothing that bites off more than he can chew

I'm working on a power amp build and I had a few questions. I will scan/upload a copy of my schematic upon request. Generally, the gist of it is I took an existing schematic and modified it. I am essentially constructing two 4-stage pre-amps (2 x 12AX7) and tying those into 2 EL84 power tubes. I later changed those to KT88s for more power.

I intend to run the 12AX7 plate supplies at 250V and the KT88s at close to 800V so I can extract maximum power.

I'm shooting to be able to drive a 70W speaker on this build.

Now here begins the questions.

1. Changing the power valve stage changes the output impedance appreciably. How would I measure the output impedance (beware, n00b ideas coming) - is it as simple as powering the valves and putting an ohmmeter across the signal path?

2. Relative to 1, would that change in impedance pose a problem when picking an appropriate output transformer? If so, would it be possible to use an L-pad to convert to the appropriate impedance for the output transformer?

Obviously, a passive L-pad would incur a power loss, and there are no active designs I can think of that are capable of operating at power valve ranges aside from well...power valves. But is the change in impedance caused by the change in valves in the power stage appreciable enough that if I were to use an L-pad to match the two, would the power loss be significant?

3. Right now, the circuit (which I will post in a day or so - as soon as finals are over I will be home and have access to a scanner) is configured as two independent single-ended single paths summed at the output transformer.

First of all, why is it that power valves exhibit higher power ratings when ran in push-pull configuration? Secondly, how would I configure two KT88s to run in push-pull?

4. Rather unrelated question, but why is it that voltage coming out of a transformer typically increases in voltage a certain amount after rectification? Was wondering because I need to account for this when assigning resistance to drop the B+ rail to acceptable plate loads for the valves.

Thanks in advance, and I will have the schematic up as soon as I have a scanner to work on.

Eli Fedele
(solidus)

 

[duke37]

1. The load on a valve is determined by the optimum impedance to get the maximum power out. It has nothing to do with the output impedance of the valve or the transformer resistance. Check up 'load line'.

2. The transformer is to transform the speaker impedance to the optimum load for the valve. No resistors should be used, these will only waste power.

3. Do not feed two outputs into one transformer, they will fight one another.
A push pull circuit uses a pair of valves where when one valve increases current, the outher decreases. By using windings appropriate windings the two outputs are added.

When in push pull, the valves sit with very little static current and only pass a lot of current on power peaks, so the peaks can be much larger than when the valves are running in class A where there a large constant current passing.

4 The voltage of a sine wave goes up and down. The voltage is normally specified as the equivalent voltage of a DC supply (RMS). In other words, sometimes the voltage is above the figure quoted and sometimes below. The peak voltage of a sine wave is Vrms * SQRT(2). When nomally fed to a rectifier, the higher voltage is selected, If a choke input filter is used, then the voltage out is the mean voltage. Vout = 0.9 * Vrms. Choke input filters are used for high powered amplifiers since it is kinder on the components.
To get the correct voltage on the valves, chose the right supply transformer, do not use resistance.

There are many circuits out there, some by Mullard, you can find them on the net.
An EL84 or 6V6 in class A will give sufficient output for domestic listening.
A pair of EL84 in class B will give sufficient power to damage domestic speakers.
A single KT88 will keep your house warm and give no more power than a pair of EL84 in class B
A pair of KT88 or EL34 will give a very large output but needs a very high voltage and complicated power supply. This has the advantage that it only kills once.

 

[CDRIVE]

Google "Maximum Power Theorem".

Chris

 

[shrtrnd]

I had to replace a few KT88's in a 1957 Taylor Hobson Test and Measurement instrument last year.
Have you checked the prices on those things?
Imagine what the price will be when you need to replace them.
I'd consider something else recommended above, if I were you.

 

[Solidus]

Sorry for the delay, got home; haven't had much time to even think aside from drawing a few plans and schematics, so on, so forth.

I'll try to address all these posts in a single shot.

I've converted the schematic (with a lot of headache and cross-referencing) to a push-pull design. Bear with me for this.

1. The load on a valve is determined by the optimum impedance to get the maximum power out. It has nothing to do with the output impedance of the valve or the transformer resistance. Check up 'load line'.

2. The transformer is to transform the speaker impedance to the optimum load for the valve. No resistors should be used, these will only waste power.

3. Do not feed two outputs into one transformer, they will fight one another.
A push pull circuit uses a pair of valves where when one valve increases current, the outher decreases. By using windings appropriate windings the two outputs are added.

When in push pull, the valves sit with very little static current and only pass a lot of current on power peaks, so the peaks can be much larger than when the valves are running in class A where there a large constant current passing.

4 The voltage of a sine wave goes up and down. The voltage is normally specified as the equivalent voltage of a DC supply (RMS). In other words, sometimes the voltage is above the figure quoted and sometimes below. The peak voltage of a sine wave is Vrms * SQRT(2). When nomally fed to a rectifier, the higher voltage is selected, If a choke input filter is used, then the voltage out is the mean voltage. Vout = 0.9 * Vrms. Choke input filters are used for high powered amplifiers since it is kinder on the components.
To get the correct voltage on the valves, chose the right supply transformer, do not use resistance.

There are many circuits out there, some by Mullard, you can find them on the net.
An EL84 or 6V6 in class A will give sufficient output for domestic listening.
A pair of EL84 in class B will give sufficient power to damage domestic speakers.
A single KT88 will keep your house warm and give no more power than a pair of EL84 in class B
A pair of KT88 or EL34 will give a very large output but needs a very high voltage and complicated power supply. This has the advantage that it only kills once.

Okay, so if I understand correctly, the two factors that change the impedance of the circuit are:

1) the input impedance (the supply)
2) the speaker connected to it (the load)

Am I on the right track?

Also, is there a method to ascertain the voltage after rectification that doesn't involve "brute-forcing" it (i.e. rectifying it manually and holding a multimeter to the live leads)? I plan on using a transformer capable of putting 450-0-450 across the mainlines so if I don't have to deal with those voltages directly I won't..

I've already put about 800V through the tip of my finger before when using an oscilloscope on the last amp build, so I've had enough share of electrocution.

The reason I intend to use KT88 valves is not a blind pursuit of output (it's not an "I want my amp to be louder than my friend's"-style argument). It's that I intend to drive a 70W speaker and from what research I've done, the only valve that can do that at a maximum of 2 of itself is the KT88.

I am perfectly willing to take the safety precautions to deal with the high voltages, including flicking the power switch with a 10-foot-long stick if need be.

This amplifier will end up costing me $300 at least to constuct so I intend to do my research and know what to expect at pretty much every facet before attempting construction or purchasing components.

I'm going to learn how to SPICE model and have a suitable workup done before this progresses out of "plan" stage.

Google "Maximum Power Theorem".

Chris

Chris; does this explain the workings of the output transformer?

I had to replace a few KT88's in a 1957 Taylor Hobson Test and Measurement instrument last year.
Have you checked the prices on those things?
Imagine what the price will be when you need to replace them.
I'd consider something else recommended above, if I were you.

I've been able to find new valves for about $60/matched pair online, so the cost of valves isn't a major deterrent at this point. Brings it down to around $30 a valve, which is about $5 more than what I've paid for top-quality 12AX7s in the past.

Not that the lowest price possible isn't a goal, but I figured better to get 2 KT88s rather than have to go 4 EL84s for the same or less power. It also helps as I'm trying to build the amp in a small enclosure (cigar box ftw).

KT88s are behemoths of valves, though.

 

[duke37]

The 450V-0-450V transformer will give you 636V when rectified. This will be above the normal electrolytic capacitor voltage, you will need two in series with balancing resistors. A choke filter should be added and then large capacitors to cater for the current pulses demanded by the class B circuit.

What do you mean by 'brute-forcing' and 'manually rectifying'?

My valve book says 550V on the anodes will give 100W.
The load on the valves should be 4500 ohms anode to anode, so, with a 15 ohm speaker the transformer ratio should be SQRT(4500/15) = 17:1
A good transformer is likely to cost a LOT of money.

There are other valves which can give 100W out, the EL34 can be used with up to 800V on the anode which will match your power transformer better. These would need an anode to anode load of 11000 ohms.
Color (colour) TV 'sweep' toobs (valves) can be used for cheap and cheerful amps, the PL509 or PL519 were used in the UK but need 30V heaters.

You will need a large enclosure to keep the temperature reasonable and it will need to be strong to carry the weight.

A pair of TY4-500 will give 2440W out with 3500V on the anodes!

 

[CDRIVE]

Google "Maximum Power Theorem".

Chris

Chris; does this explain the workings of the output transformer?

No, it dictates a basic law, just as Ohms Law does. The output impedance of your amplifier is not governed by the output transformer alone. Your finals have an output impedance too.

Following 'Maximum Power Theorem' maximum power will be transfered from the plates to the speaker load when the following are satisfied:

(A) When the speaker impedance matches the output impedance of the output transformer.
(B) When the input impedance of the output transformer matches the output impedance of the finals.

Note: The rated output impedance of an output transformer (A) is only valid when (B) is satisfied.


Chris

 

[duke37]

Chris is strictly correct to get maximum power but this is not what you want. If you go for maximum power, you will likely burn out the valves.

For an audio amplifier, there is an optimum load line, which approximates to a line from minimum votage and maximum current to maximum voltage and minimum current. This is quoted in ohms and is given in valve data.

The output impedance of a tetrode will be much higher than the load impedance and, in the case of a pentode could be over 1M.

 

[Solidus]

The 450V-0-450V transformer will give you 636V when rectified. This will be above the normal electrolytic capacitor voltage, you will need two in series with balancing resistors. A choke filter should be added and then large capacitors to cater for the current pulses demanded by the class B circuit.

For future reference, how did you calculate that, or did you simulate it?

How would I arrange that circuit? I'm questionable on the idea of balancing resistors. Are those in series with the capacitors?

What do you mean by 'brute-forcing' and 'manually rectifying'?

My apologies for the ambiguity. I meant I'd rather simulate the circuit and know what to expect beforehand rather than solder rectification diodes on and then manually hold a multimeter to the live wire tips. Not with 636V.

My valve book says 550V on the anodes will give 100W.
The load on the valves should be 4500 ohms anode to anode, so, with a 15 ohm speaker the transformer ratio should be SQRT(4500/15) = 17:1
A good transformer is likely to cost a LOT of money.

There are other valves which can give 100W out, the EL34 can be used with up to 800V on the anode which will match your power transformer better. These would need an anode to anode load of 11000 ohms.

Will a single EL34 push-pull pair reach 100W? I'm not opposed to using them, I just didn't believe they could reach those power levels without grossly overdoing plate voltage.

A pair of TY4-500 will give 2440W out with 3500V on the anodes!

I'm trying to build a guitar amplifier, not demolish a wall! haha

Out of curiosity, I checked the datasheet, that bad boy of a valve requires -350V at the grid, consumes just under 4kV at the plate and can be used up to 120 MHz.

I'll remember this if I ever try to make a pirate radio station.

 

[duke37]

To get the peak voltage, multiply the RMS voltage by the square root of 2.
450 * 1.414 = 636

This needs to be fed into a moderate sized capacitor, a very large one will cause the current to flow on just the peaks of the voltage and this high current will cause overheating of the transformer. You will get a somewhat lower voltage but more ripple.

This ripple can be smoothed with a choke and another capacitor. A class B amplifier uses little current with little output but takes a lot of current at high output. The capacitor needs to be big enough to supply these surges.

If you simulate the power supply cicuit, you will need to provide a fluctuating load.

Two capacitors in series will be needed to get sufficient voltage rating and if they are identical, that is all that is necessary. However, no two capacitors are identical and one will have a higher leakage current and so the voltage will not be equally shared. Adding a resistor across each capacitor will even the voltage but at the expense of power loss.

I have not looked up the full details of the EL34 but my Iliffe valve book gives the conditions to get 100W from a pair. If you go for 70W instead of 100, then you will not notice the difference and the component will last longer.

 

[Solidus]

What are the conditions to push 70W?

 

[duke37]

375 V and 107mA will give 48W
800 v and 91mA will give 100W
600 V will therefore give about 70W. Load varies from 3500 to 11000 ohms so chose about 5000 ohms.

You will need to set the bias, drive and load lines for your particular situation.

 

[Solidus]

I will have the schematic up tomorrow sometime, brain is fried from Christmas shopping haha.

On a practical note, how would I go about measuring impedance?

 

[duke37]

The impedance of speakers vary wildly depending on the frequency and the box in which they are installed. Just use the value marked on them.

Happy Christmas

 

[Solidus]

The impedance of speakers vary wildly depending on the frequency and the box in which they are installed. Just use the value marked on them.

Happy Christmas

Well, I was more referring to the actual amplifier circuit.

A few of my friends have expressed interest in my amplifier-building should I get to some proficiency with it (money! ), and I enjoy the designing and building enough to consider doing it as a side hobby, so if possible, I'd like to accrue (or build) equipment capable of testing the circuits in various regards.

Also, while on the tangent, and without delving into an essay of posts discussing major factors, what determines impedance? Correlating one of your posts, it (in valve amps) seems to be proportional to the plate voltage (and therefore current) and as you stated above, frequency has a component.

From having read up on it to a significant degree (and not having much luck understanding), impedance is the complex AC analogue to resistance. Now, how is impedance different then just measuring the resistance of the circuit at a particular frequency?

I also understand the notion of reactance factors in. I've looked at reactance in capacitors; albeit not thoroughly - I understand that it in a nutshell is the resistance of a circuit to a fluctuating current or signal due to capacitance or inductance.

(A bit of this will be out-loud thinking; let me know if I'm on the right track)

If the reactance is inversely related to frequency, a low frequency has a high reactance, and if Z is proportional (I know it factors in a complex variable, so I'm cutting that out and making a simple proportion) to the the sum of the resistance R and the reactance X, a large reactance creates a high impedance which attenuates low-frequency portions of signals. Is this correct?

 

[duke37]

I would suggest that you use a standard circuit especially if you are going into business.

The required load impedance of the amplifier is determined by the load line and is given in valve data. The output impedance of the valves could be many times the load impedance and there is no point in measuring it.

The matching of the various stages in the amplifier is fairly simple, if there is a series capacitor just make it large enough to pass the lowest frequency without too much attenuation. Too large a capacitor can give trouble from unwanted feedback through the power supply.

To smooth the power supply, in general, the larger the capacitor the better.

If you are going to use feedback to reduce distortion, then things can get very difficult since phase changes can occur in any stage.

 

[Solidus]

I dunno about the business aspect of it. Depends if I am able to show proficiency at construction and whether I retain the interest to construct the amplifiers.

When designing the circuits, I'm using pretty standard valve topologies - single-ended, triode-strapped, cathode follower, push-pull. Not trying to reinvent the wheel. That's for when I have the money/time/resources to design, build, test, rework, etc. and I'm not pressed for getting a project done in minimal resources.

Would 3x100uF caps shorted to ground be sufficient to smooth the B+ plate lines?

I don't intend to use feedback. The simplicity is nice, but one of the main reasons for using valves over solid-state is the fact that the distortion character is better, and so instrument distortion is welcomed in valve amps, if not deliberately induced.