Negative voltages?
I am analysing the attached circuit to determine the sign of the voltage at the input, and at the large feedback capacitor terminal that is not connected to ground. LTSpice is telling me the small voltages are NEGATIVE with respect to ground. This means, the large capacitor should be soldered with its cathode to the resistor and the anode to ground. The input capacitor should be connected with its cathode to the base of the input transistor.
Am I right?
I am analysing the attached circuit to determine the sign of the voltage at the input, and at the large feedback capacitor terminal that is not connected to ground. LTSpice is telling me the small voltages are NEGATIVE with respect to ground. This means, the large capacitor should be soldered with its cathode to the resistor and the anode to ground. The input capacitor should be connected with its cathode to the base of the input transistor.
Am I right?
Three more transistors to complete the circuit. Tomorrow, hopefully, I should be able to finish interconnecting the input, VAS, VBE multiplier and overcurrent limiter PCB with the main PCB which hosts the output stage and smoothing capacitors.
After that, it will be time to do testing with audio signals.
After that, it will be time to do testing with audio signals.
The remaining parts arrived. I will try not to let excitement direct me. Being the last phase before trying the amplifier with an audio signal, this phase is critical and errors would only mean blown parts that take weeks to arrive by post.
I will try to be cautious and very patient.
When I power the complete amplifier circuit with a feeble power supply, I will test various critical points to verify the DC voltages are as they should.
I will try to be cautious and very patient.
When I power the complete amplifier circuit with a feeble power supply, I will test various critical points to verify the DC voltages are as they should.
First Test Results:
Quiescent Voltage at output: 8.5mV
Voltage at Base of Non-Inverting Differential Transistor: -45.2mV
Voltage at Base of Inverting Differential Transistor: -46.8mV
Votage at Collector of VAS's Amplifying Transistor (PNP): 0.65V
Voltage at Collector of VAS's Current Mirror: -0.12V
Input Stage Constant Current: 3.1mA
VAS's Standing Constant Current: 15.7mA
Putting a very high resistance between the non-inverting input and +Ve and -Ve Rails forced the output voltage to clamp at the positive rail and negative rail respectively.
The above shows it is time to connect a speaker and an input signal.
Err. No! The voltage across the VBE multiplier is far too low.
Quiescent Voltage at output: 8.5mV
Voltage at Base of Non-Inverting Differential Transistor: -45.2mV
Voltage at Base of Inverting Differential Transistor: -46.8mV
Votage at Collector of VAS's Amplifying Transistor (PNP): 0.65V
Voltage at Collector of VAS's Current Mirror: -0.12V
Input Stage Constant Current: 3.1mA
VAS's Standing Constant Current: 15.7mA
Putting a very high resistance between the non-inverting input and +Ve and -Ve Rails forced the output voltage to clamp at the positive rail and negative rail respectively.
The above shows it is time to connect a speaker and an input signal.
Err. No! The voltage across the VBE multiplier is far too low.
Last edited:
Ooops! VBE transistor connected the wrong way!
The VBE multiplier transistor was connected the wrong way. Connecting it the right way and testing it was not blown, although the currents used so far are too low to do any damage, resulted in more current drawn from the power supplies. The VBE voltage rose to 2.25V.
So far, it is clear the 50 Ohm preset is not enough. I need another one with a higher resistance. However, changing values for this preset resulted in the power supply voltages rising to 5V from around 3.5V.
The power supply used so far is a very weak power supply that can supply +/-15V DC at a current of around 400mA.
The VBE multiplier transistor was connected the wrong way. Connecting it the right way and testing it was not blown, although the currents used so far are too low to do any damage, resulted in more current drawn from the power supplies. The VBE voltage rose to 2.25V.
So far, it is clear the 50 Ohm preset is not enough. I need another one with a higher resistance. However, changing values for this preset resulted in the power supply voltages rising to 5V from around 3.5V.
The power supply used so far is a very weak power supply that can supply +/-15V DC at a current of around 400mA.
Wow, what sensitivity! Without a cable soldered to the input, the amplifier can 'feel' the presence of my index finger from a few millimeters above the PCB where there are the input tracks.
Trying the amplifier with an audio signal source it worked, but there is still an issue with the VBE multiplier. Therefore, the output is not optimal. Observation indicates the base-emitter resistance network is not being used. Probably, this is due to a bad solder joint. I used a 10 turn micro preset from the original circuit but the pins were too short.
I would like to ask, what kind of presets are normally used for the VBE multiplier? Is it enough to use a normal preset; id est, those that can be turned to about three quarters of a revolution?
Trying the amplifier with an audio signal source it worked, but there is still an issue with the VBE multiplier. Therefore, the output is not optimal. Observation indicates the base-emitter resistance network is not being used. Probably, this is due to a bad solder joint. I used a 10 turn micro preset from the original circuit but the pins were too short.
I would like to ask, what kind of presets are normally used for the VBE multiplier? Is it enough to use a normal preset; id est, those that can be turned to about three quarters of a revolution?
The sub-optimal sound is due to a very low VBE bias. The total current drawn under no signal conditions per supply rail should be 95mA, but measurement of this same current is only showing 20mA.
The currents drawn are as follows:
a) input stage: 3mA
b) VAS: 15mA
c) remainder parts 5mA
The remaining parts should be drawing far more current for the output stage to be in optimal bias. The output bases are fed from across a 22 Ohm resistance. This means, for the voltage to reach conduction threshold for the base-emitter junction (of the output stage) a current of 0.50V/22 Ohm = 22.7mA is necessary. Adding to this 12mA*4 devices gives another 48mA.
Therefore:
i) input current: 3mA
ii) VAS current: 15mA
iii) Output stage driver: 22.7mA
iv) Output stage bias current: 48mA
This gives a current of: 88.7mA which is very close to what LTSpice is calculating.
The currents drawn are as follows:
a) input stage: 3mA
b) VAS: 15mA
c) remainder parts 5mA
The remaining parts should be drawing far more current for the output stage to be in optimal bias. The output bases are fed from across a 22 Ohm resistance. This means, for the voltage to reach conduction threshold for the base-emitter junction (of the output stage) a current of 0.50V/22 Ohm = 22.7mA is necessary. Adding to this 12mA*4 devices gives another 48mA.
Therefore:
i) input current: 3mA
ii) VAS current: 15mA
iii) Output stage driver: 22.7mA
iv) Output stage bias current: 48mA
This gives a current of: 88.7mA which is very close to what LTSpice is calculating.
That could be a sign of instability. I can't remember what the input impedance looked like but it should be low enough to rule out true stray pickup that you might get from a FET based stage of many meg ohms input Z.
Might be worth looking at adding an RF filter at the input or even slipping a ferrite bead over the base lead of the input transistors.
Its good to know it all seems otherwise basically OK though.
Presets are invariably the small types commonly available, they have been used for decades and don't really seem to be a problem. Just use a decent quality one, then 'set and forget'.
It is pickup of a hum signal going to earth. It was a feeble audible mains hum the strength of which, suggests there is no problem in the input stage. In fact, connecting a coaxial signal cable to the input, resolved the issue, as the stray signal is shunted to earth through the parasitic capacitance of the cable. The input already has a low pass filter.
I am uncertain about available presets quality. The ones I remember from over thirty years ago where far more robust. Plastic pakaging gives me a sense of uncertainty about their quality.
For a Miller capacitance I need a stable 100pF capacitor that can withstand operating with a power supply of 170V. I tried to find a polyester or tantalum capacitors with these specifications and failed to find any.
Ceramic capacitors seem to satify both capacitance and voltage requirements, but they seem to have a high tolerance (+/-20%). This means a possible uncertainty range of 40%.
Ceramic capacitors seem to satify both capacitance and voltage requirements, but they seem to have a high tolerance (+/-20%). This means a possible uncertainty range of 40%.
Plenty out there I think:
https://uk.farnell.com/lcr-components/fscex-100pf-1-630v/cap-100pf-630v-1-ps-through-hole/dp/9520660
https://uk.farnell.com/lcr-components/fscex-100pf-1-630v/cap-100pf-630v-1-ps-through-hole/dp/9520660
Sticking point
If you are still using either of these supplies then you will have reached a sticking point with your tests and you will need to find a way to increase the rail voltages while restricting the amount of current you can draw from these.
Where you are at with testing has led to a suggestion there is some presence of instability. Pursuing this could be time wasting in the foregoing set up.
Assuming no change, it might help to understand one possible cause if it is understood that collector to base capacitance in a common emitter stage is relevant and the LTP and Vas stages are cases in point.
The collector to base capacitance is an indication of the width of the collector and base diffusions with the former being somewhat elastic depending on the voltage applied at the collector. If this is caused to expand the base will contract somewhat and there will be a small loss of current gain. (Read about Early Effect)
At the same time a common emitter amplifier being an inverting stage will amplify the collector to base capacitance creating an off target load at the base.
These things will be in better balance if the collector voltage is high as the collector to base capacitance has an inverse relationship with the collector voltage and a direct one with the area of the junction - so in approximate terms the collector to base capacitance is inversely proportionate to the square root of the collector voltage.
Plucking some values out of the air 15V and 70V the roots are 3.87 and 8.36. From this the collector to base capacitance at 15V will be roughly double the value at 70V.
In these circumstances your LTP would have double the input capacitance to drive at the input of your Vas (Miller Effect). All this added complication can be side-stepped by increasing the supply rail voltages.
While you could try a lower wattage light bulb to allow a little more current flow to allow your Vbe adjustment to work and increase the rail voltages a little, I think it would be better to work out what current limit you want to set and calculate values for resistors by Ohms law to fit in the dc supply rails instead.
This method has been used for projects of various power outputs in Australian Electronics magazines for over 40 years and used in thousands of kit sets built in that time.
If you are still using either of these supplies then you will have reached a sticking point with your tests and you will need to find a way to increase the rail voltages while restricting the amount of current you can draw from these.
Where you are at with testing has led to a suggestion there is some presence of instability. Pursuing this could be time wasting in the foregoing set up.
Assuming no change, it might help to understand one possible cause if it is understood that collector to base capacitance in a common emitter stage is relevant and the LTP and Vas stages are cases in point.
The collector to base capacitance is an indication of the width of the collector and base diffusions with the former being somewhat elastic depending on the voltage applied at the collector. If this is caused to expand the base will contract somewhat and there will be a small loss of current gain. (Read about Early Effect)
At the same time a common emitter amplifier being an inverting stage will amplify the collector to base capacitance creating an off target load at the base.
These things will be in better balance if the collector voltage is high as the collector to base capacitance has an inverse relationship with the collector voltage and a direct one with the area of the junction - so in approximate terms the collector to base capacitance is inversely proportionate to the square root of the collector voltage.
Plucking some values out of the air 15V and 70V the roots are 3.87 and 8.36. From this the collector to base capacitance at 15V will be roughly double the value at 70V.
In these circumstances your LTP would have double the input capacitance to drive at the input of your Vas (Miller Effect). All this added complication can be side-stepped by increasing the supply rail voltages.
While you could try a lower wattage light bulb to allow a little more current flow to allow your Vbe adjustment to work and increase the rail voltages a little, I think it would be better to work out what current limit you want to set and calculate values for resistors by Ohms law to fit in the dc supply rails instead.
This method has been used for projects of various power outputs in Australian Electronics magazines for over 40 years and used in thousands of kit sets built in that time.
I looked up a write up of a 350W amplifier published by Silicon Chip magazine in 2004 this ran from 70 volt supply rails. They specified a 250 volt 68pF ceramic disc (or mica) capacitor in the Vas Miller position (a BF469) and provided a link to Farnell part 867-871. The suggested replacement in 2019 is a 500V silver mica 1% type with compact lead spacing from the CD15 series of product.
It is a good idea to keep the Miller cap leads and collector and base Vas transistor traces as short as you can manage. The use of resistors in the dc supply connections to the amplifer board method was used in this project and these were 470 Ohm 10 W rated so with 85 volt rails you would need to increase the resistance to 560 Ohms to get the same level of protection - there are four MJL21193/94 power transistor per module sets in this project.
With 85 volt power supplies there will be a lot more energy in your supply capacitors and in the article it was recommended that these be covered by a rigid non- conductive shield - such as a sheet of clear perspex in view risks of fatalities from accidental personal contact with +/- 70 volt supplies ( increased in the case of +/-85 volt ones). This protection was recommended for attention before the amplifier was powered on.
The use of 10A cable for all power and speaker connections was specified with ends terminated with 6.3mm push on connectors to suit board mounting lugs - just to mention some of the other construction detail.
Last edited:
There is an alternative for the cap solution. Two capacitors can be used in series. The center point can be tied to a half voltage point by a high value resistor. A second alternative is to use a much bigger value in series with the real value you want to implement. In this case tie the center point to a lower voltage because the smaller cap has lower voltage specs. The tie resistor reduces the voltage requirement and having two in series helps reduce overall leakage. You will need to factor in the charging current at power up so that the lower voltage cap specs are not exceeded. This technique works with valve amps.