PWM amplifier 1 - The PWM modulator design I.

Newly detected TAS5630 chip can solve my problems about the PWM amplifier. This chip have analog inputs, the package contains all required parts (A/D converter, PWM modulator), and the parallel mode is possible if less than 4 Ohm output impedance required. with this package we have one paralleled mono or one stereo poweramp with analog input.

But I have TAS5261 chips, without analog input. For this design required TAS55xx PWM modulators and analog/digital converters. These chips recommended for home DVD  players, the power supply is 3.3 volts only. I don't think that this solution is ideal for instrument amplification.

In this post, I writing about the method and simulation of analog PWM modulator.

First I need a triangle function generator. I found too much... The clock of PWM input TAS chip is 192-384 kHz, what is understandable if digital chips placed before chip. For this frequency, not too easy to find triangle signal generator with very good output.

Can be found one-chip function generators. One of them is LM566 (with this chip I found a schematic) not available here and now, or we can use XR2206 or ICL8038 chips. These chips are very simple solutions for triangle output, maybe I will try some of them. The maximum frequency on the datasheet is 0.5MHz, typical is 1MHz, but the 8038 have 300kHz maximum. I have no experiences with these function generator chips, so I don't know about the output quality. This is the reason, why I seek discreet analog function generator for triangle output. The another reason, is, function generator chips are not available on Multisim.


The first tryouts I have no success:

This function generator promised for perfect output, but the triangle with 20 kHz only is really bad. So I don't need to try this design for 200-300 kHz.

I tried more triangle generators with op-amps, but the output is not good enough. Finally I found a really good analog triangle generator with nice triangle and square outputs. This circuit can be used up to 4MHz what is more than enough.

I have good result on 300kHz:

See the simulation output, this is 235 kHz. Very easy to set the frequency and duty cycle.

I tested this discreet (but easy to set) schematic  on higher frequency. Here ate the test results:

1.3 MHz:

2MHz:

3.5 MHz:

6 MHz:

I think, if the upper schematic is good enough on 2-4 MHz, then this is good for around 300 kHz only.

The triangle output of this circuit wired to the fast comparator. the most of comparators doesn't worked on Multisim for me, but finally I found one what is very good for output simulation:
 

Example about the signals in the circuit:

The purple line is the triangle output of function generator, the green is the part of 1kHz sinus (audio) input, the red is the TTL PWM modulated square with variable duty cycle.

In this image can be seen how duty cycle modulated by the sinus input signal:

Continued the circuit simulation, the next is the comparator. This part AD8611 is good on Multisim, but I think another comparator (for example LM311) is good on real life too. The triangle signal and the audio (sinus) input have to be connected to the comparator's inputs.

The most important question is, the quality of the PWM modulated output. What I get back, if the PWM modulated square signal converted back to the original sinus wave. The quality of this method cannot seen by the quality of the PWM modulated output. See the image:

The green line is the original audio sine wave signal. I have to get back this wave from the PWM outputs. The dark-red is the (good) result if the PWM output converted back to the original sine. What is the blue line, I will tell you soon....

After the comparator chip, I put TTL logic. This logic made soft start while the triangle generator up, and disabled the input of TAS H-bridge. This method made inverted and non-inverted PWM outputs. Finally this is the "deat time" logic. The TAS5261 contains "dead time" logic, and the most if comparators have inverted and non-inverted outputs. This TTL logic required, if the output of PWM amplifier is simple MOSFETs, not H-bridge chips with internal logic what I want to use.

......the blue line (what is really wrong and distorted) is the converted output of TTL chips.

Here are the signals in this design: The sine is 20 kHz now, the blue line is the triangle signal on 235 kHz, the red line is the TTL PWM output, the tall brown line is the output of comparator. The result is much better on the comparator output.

After simulation results I have many problems.

The first question is about the precision of simulation. How about the simulation and the  reality, whats about the differences, depending on the PCB design and the quality of parts.

After these PWM modulated square waves, I have to put TAS H-bridhe chips, and I can use these outputs for audio. Maybe this method have good result, but maybe not. Maybe the TTL outputs (what is really wrong after decoding on the simulation result) good enough for discreet MOSFET drivers. Finally I have to test the DC offset of comparator inputs and outputs, and the linear frequency range of the decoded output. These solutions is on the next post.

Prototype PCB with soldering mask

I made this movie how I made very good quality prototype PCB with soldering mask.

New PWM power chips with analog inputs

In the previous article I thinking about the poweramps of my guitar/vocal amplifier. The power supply is ready, I have 1200VA, this is the reason why I want to include more than one guitar amp to this new design. But whats about the top of this amplifier case, where I need poweramps and preamps?

I think for the guitar amp I will use bridged TDA7293 because for 8 or 16 Ohm speakers the bridged amp solution is better. But for keyboard/vocal I wanted to use Class-D TAS chips. These chips have 125W output power for 8 Ohm, but 315W for 4 Ohm. The impedance of additional cabinet for vocal or for guitar can be 4 Ohm (or less) if required. In my previous DIY amplifier I used bridged TDA chips, and I really like it. This is the reason why I want "hybrid" solution. One analog amplifier for guitar with TDA chips, and another one for all others with PWM solution.


This is one example that the 4x200W TDA bridged solution is more than enough for clubs for guitar and vocal (but this is gym):



For the PWM solution I think about TAS5261 chips. This is small power bridge with PWM input, but PWM modulator and A/D converter required. I have no experience with these chips, and these solutions have very small size, I am not sure that I can assemble them.

This chip is 2x210W PWM power-bridge with the smaller package.

This is TAS5261 chip, 1x350W power bridge.

I like these solutions, error and speaker protections included, chips have very good efficiency, no large heatshink required. The efficiency of TDA solution is about 60%, but his PWM chip have more than 90%.

But these TAS chips offered for home theater or home DVD systems, maybe not for instrument amplification. See the datasheet:

• 315W - 4 Ohm
• 220W - 6 Ohm
• 125W - 8 Ohm

This is the reason why I don't want to use this solution for 8 Ohm guitar speakers, but maybe want to use for 4 Ohm vocal or additional cabinets.


The another thing that I don't want to use required additional circuits like A/D converters and PWM modulators. These chips have very small package, and the power is 3.3V only.

The PWM modulator:

The one pin of this chip 0.2mm width, the distance between the middle line of pins are 0.5mm. Very hard to solder and assemble these parts.


But now the TAS PWM power bridges have analog inputs. PWM modulators and A/D converters not required. The package of these new solutions have 44 pin DKD, and the new 64 pin PHD. The 44 pin DKD package is same as like the previous TAS5261 chip. The biggest version of new analog input TAS chips have 600W power output, what is really great on very small size. See the datasheet page 18.

The benefits:

• Very good efficiency (85...93%) no large heatshink (and cooler) required
• Have analog inputs, no PWM modulators and A/D converters needed
• The chip have error protection

But:

• The 600W available on on 2 Ohm cabinet (the output of chips are paralleled)
• 4 Ohm - 300W, what is not bad, but the previous chip have same feature.
• 3 Ohm - 400W (paralleled outputs)
• 6 Ohm - 210W
• 8 Ohm - 160W.

But with these output powers have distortion 10% what is not small enough. If we need better quality, max. 1% distortion, the possible power will much smaller, the chip cannot be paralleled:

• 4 Ohm - 240W. This is same as TDA bridged amplifier but on 8 Ohm impedance.
• 6 Ohm - 160W
• 8 Ohm - 125W

With smaller impedance with paralleled outputs on 1% distortion:

• 3 Ohm - 310W
• 2 Ohm - 480W

I found these solutions if I want to use PWM amplifier. The PWM chip is better than older solutions if I have speaker cabinet with 4 Ohm or less. I am not decided yet what will the poweramp on the top case of my guitar combo:

About the top of my new guitar amplifier design

This project stopped for years, but the wood case of my new guitar amplifier is ready. The rason of long time delay while I did nothing is that I used my old amplifiers. One is my previous DIY guitar amp have been disasembled, and I have a Carlsbro GLX100 amplifier with Eminence ReadCoat speaker, whats are very good solutions. This is the reason why I don't hurry with my new idea, but now I take apart my old DIY amplifier and the power supply have been moved to the new wood-case.

My old rack mounted amplifier was stereo for guitar, because I  using multieffect. I have Boss GT8, where not only the effects have stereo outputs, this is dual guitar effect, where I can use different amplifiers for left and right channels. This is the reason, why I have combo design but with two speakers for stereo solution:

After the wood case, I have metal cases for electronic circuits:

This is the back of the wood case:

....now without the middle slat because I need space to fix the speakers.

 The power supply built to the bottom of wood-case. I know that all other industrial solution have same case with other circuits on the top, but now I have more space on the top for amplifiers and equalizers. I think the larger distance between transformer and circuits have benefits:

The design of wood case was realy good, I made model of paper, and got to the professional carpenter, who manufactured and assembled:

The bottom of this amplifier is done, this is the power supply. From the power supply the required voltages wired to the top of amplifier case:

Here I have +/- 40V for poweramps, +/-18V for preamps, +24V for LED-s relays, and coolers.

But the most important question is what I have on the top

Because the power supply built to the bottom, I have more space on the top. The power of toroid transformer is 1200VA. I want to place guitar preamplifiers and amplifier (stereo for multieffect), and one more stereo poweramp for additional cabinets for vocal or keyboards. I need parametric equalizer, if I have enough space on the top-case, separated for left and right channels. I need audio mixer with mic preamps for vocal or another audio sources with headphone and line outputs.


But the most important question is, what will be the power amplifier on the top case. On this time, I want to build some digital class-D poweramp maybe for vocal only not for guitar. I have PWM power bridges, what is 1x315W, and I have 2x210W TAS chips:

1x350W with TAS5261 is would be enough for guitar, but this power available for 4 Ohm only. The guitar speakers are 8 or 16 Ohms. On 8 Ohm this chip have 125W output power. Now I think, for guitar (where the speaker is 8 Ohm) I will use bridged TDA7293 chips, where I got 200-220W for 8 Ohm speakers. This chip is very popular on guitar amplifier market, Marshall, Carlsbro using this solution. And for example the Crate Powerblock is PWM poweramp.


Because the vocal or additional guitar speakers have less than 8 Ohm impedance, the PWM chip is good solution. The benefit of PWM chip is the very high efficiency. But the disadvantage is the very small size of chip, TAS5261 have digital PWM inputs only, so I need PWM modulator like TAS5028 for example. TAS5028 is more small than TAS5261, so more harder to build the PCB. Finally the PWM modulators have digital inputs, so I need A/D converter for analog input devices.

I'm not sure that I can build complex design with these very small chips. The another detriment is that the PWM modulator and the A/D converter need 3.3V only, what is used for DVD players orfor computer soundcards. I don't think that the vocal or guitar amplifier design is same as like the home theater systems. The power stage of this idea need only non-symmetrical power voltage, +50V. I have +/-40V for TDA bridged solution, what is much better. I can bridged the TDA chips, but I can't bridge TAS power chips, because this is already internally bridged. At he moment I think the bridged TDA circuits are better and easier to build for guitar and vocal.

I take apart my old amplifier, old circuits for sale

I "disassembled" my good old power amplifier. This version was very good but the part of this device needed to my new idea. I used this old version for 10 years, without problem. This amplifier contains two stereo devices, one for guitar, and one for microphones/keyboards. I built this double-stereo amp to 2 rack mounted cases: one for power supply, and one for the amplifiers.

The old power supply:

It was the old power supply on the rack case #1. The toroid transformer is  1200VA, the Greatz on the heatshink is 10A, softstart, stabilized power supply for preamplifiers and EQ, simple power supply for amplifiers. By the speakon connectors the output voltage can be changed from +/-10v to +/- 70V for poweramps only. One speakon is the output for poweramps, and one 5 pin connector is for preamplifiers and mixers. The toroid transformer have 4x10W and 4X15V output AC voltages for poweramps, and 2x15V for preamps, and 1x18V for LED, relays, and coolers.

This power supply is good solution for my new guitar amp project, but without speakons, because the different output voltages are not required for the new design. Now I have only one speakon for poweramps, and one 5 pin connector for all others:

This is the box of new power supply, assembled from the parts of old one. This case placed to the bottom of new combo. The required power voltages will be wired to the top of amplifier. The content of upper case not ready, and I want to design something new.

The old amplifier:

The poweramp, 6 input mixer,10 channel equalizer built to another rack mounted case. This is old photo with one poweramp on heatshink, but this part of amplifier has been duplicated. One of then used for guitar, the another one is for keyboard and vocal. This is all-in-one sulution, but for small or larger clubs was good solution:

There are the list of very good circuits what I don't  need for new amplifier design, so there are for sale:

4x200W poweramps with TDA7293 assembled to heatshink with cooler:

The poweramps are two stereo circuits with bridged TDA7293/7294 circuits. Very reliable circuits. All stereo boards have speaker protection with UPC1237 chip.

The equalizer:

10 bands stereo RANE copy equalizer. This is active filter circuit design, the "Q" of EQ is permanent for all bands, what is very interesting solution. See more... With small PCB relays the have true bypass, and the circuit can be switch between +/-6 and +/-12 dB modes.

6 input audio mixer:

Very simple audio mixer with 6 stereo inputs. All inputs have gain and volume pots. Two of them have jFET preamplifiers for guitars with 4x gain, 4 of them have op-amp preamplifiers for all others with 100x gain.

Here are the gallery of my "disasembled" but error-free circuits and prototypes. Write me if you need something:

The power supply moved to new amp

The power supply of my stereo guitar amp is moved to the new place and ready. The old circuits can be fixed to the new "box", what is smaller but higher. The toroid transformer is 1200VA, with 4x15V, 4x10V AC for high power circuits, 2x15V AC for small power circuits, 1x18V AC for relay and LED. The DC and 24V AC voltages wired to the top of combo with 5 point cable, the +/-41V DC power for amplifier wired with speakon connector to the top of amplifier case:

Homemade PCB with soldering mask

New PCB boards made with new method. New boards have soldering mask. 

Images:

4 input headphone amplifier with TPA6120 chip

I already have schematic and PCB design with TPA6120 chip with 2 inputs. Now I made same design but with 4 inputs and one line output. The new design contains PCB transformer, dual +/- 15V power supply, 4 channel mixer without preamplifiers, unbalanced/balanced converters, and line output with level adjustment.

The schematic:


The device contains 4 audio inputs with volume adjustment. The input connector must be soldered to the bottom of PCB (blue) and the volume must be potentiometers soldered to the top (yellow) of PCB. The input and line output connectors can be 6.3mm jack OR stereo RCA connector. The headphone output can be 6.3mm jack only. The PCB contains header for 4 power filter circuits (whats are optional if not required) for 4 dual operational amplifiers. The operational amplifier chips are compatible with TL072 for cheapest solution, NE5532 for better quality, and with LT1124 for the best quality. These chips used for the input mixer, for unbalanced/balanced converters, and for the line output amplifiers.

The PCB design:


See also:

• More unbalanced to balanced converters
• Simulation of unbalanced-balanced converter
• 2 input headphone amplifier
• Headphone amplifier with TPA6120
  

PCB sales of this project Module name Size (mm) Area (cm2) PDF SCH PCB image Tested Price (US$) Des. Sim.1 Full2 Sim.1 Full2 Man3 4 input headphone amplifier 189x112 212 No Yes Yes - - No 30 39 Ask How to order? Please read the rules carefully!

More unbalanced to balanced converters

At my previous post I simulated the unbalanced/balanced converter I using for unbalanced microphone inputs and headphone amplifier. The used chips (INA217 and TPA6120) have symmetrical inputs. These chips with balanced source have much better dynamics, noise, and sound quality. I think the selected converter I used is very good, but I wanted to simulate another solutions just for fun.

1.
This very simple schematic without required capacitors. This is the reason why the AC analysis have very good result, but with required parts the result of stability and the phases would be changed:


The AC analysis of the pure schematic:


2.
This sample is more complex, but the result is not the best. The reason is that the operation amplifiers not same with inverted and non-inverted mode, but the schematic is symmetrical for - and + outputs:


The result of AC analysis:


3.
Simple "pure" solution again without capacitors and another required parts. This "base" circuit working well but with whole system have to be modified:


The AC analysis is perfect, but we got another result with required parts:


4.
On the previous 3 examples have different method negative feedback on operational amplifier. Sometime the input connected to inverted and non-inverted inputs, and the required amplifiers have several feedback. By this example the inverter circuit getting the input signal from the output of first stage. Very stable example with really good AC analysis result:


The AC analysis:


5.
The final example have more than one versions. The benefit of these schematics are the symmetrical solution of + and - outputs, which have same (or very closed) frequency responses and phases. The previous circuits have no capacitors on the negative feedback, or if have the capacitors modified the inverted stage only, the original signal more linear than the inverted. With the current solution, the frequency response and the phase of inverted and non-inverted stages are relative parallel, not like in the 2nd example what is serious problem.

The simplest version:


The AC analysis where the frequency responses are same, the phases are not linear, but running parallel, and the difference is very closed to required 180 degree:


The first modification is the active feedback between the outputs and inverted input for the adjust of inverted stage output level by the R15 resistor:


The final version of this really interesting converter is two operational amplifiers added for the output. This modification have same AC result than the previous version:


The AC analysis:


This example have gain from 707mV to 11V what is not required. If this gain is too much, modify R4/R5 and R9/R11 resistors to adjust the output gain. When the gain of stages modified, set again the output amplitude to exactly same by R15 resistor what can be trimmer potentiometer:


The result of AC analysis is very closed but modified when the gain changed:


The question is, what is the best solution if unbalanced/balanced converter required. The most simple versions are looks like perfect solutions, but with additional (and required) capacitors and with non exactly same resistors the phase and the amplitudes has been shifted. The another reason of differences is the difference between inverted and non-inverted mode of same operational amplifier stage. The simplest (and the best) schematics are block diagrams only, very good base but have to be modified. The useful solutions have two useful version: the input signal connected to same inverted and non-inverted stage. The another one is the input signal connected to non-inverted stage, and the output of this stage connected to inverted-mode operational amplifier. The another differences between the negative feedback. The second example, where the original input signal uses same stage but one with inverted one with non-inverted mode, the result is not really useful. This is the difference between modes of operational amplifiers.

The another question is the same frequency response of - and + outputs. If the original signal uses the simplest solution with direct negative feedback (like in the 1st example), and the inverted stage have negative feedback with capacitor, the frequency response (and the phases) will be different. If the difference shows after 40kHz by AC analysis, I think the result is very useful.

The last 5th example is interesting only. Not the simplest solution, and the problem is, the phase modified by the input frequency. The difference between + and - outputs is constant 180 degree, but always modified the phase values. This is the reason why I think the one of the best result for balanced conversion is on my previous post.

See also:

• Simulation of favorite unbalanced/balanced converter
• 4 input headphone amplifier
• 2 input headphone amplifier
• Modifier microphone preamplifier
  

Simulation of unbalanced - balanced converter

I using unbalanced/balanced converters with my circuits like microphone preamplifiers and headphone amp because the used chips (INA217 and TPA6120) have symmetrical balanced inputs. This is very important, the sound and the dynamics are much better with this solution. I simulated the schematic of converter, and here are the results:

With scope and AC voltmeter on real time simulation:


AC analysis:


See also:

• Simulation of several unbalanced/balanced converters
  
• 4 input headphone amplifier
• 2 input headphone amplifier
• Modifier microphone preamplifier