A75 front-end collapsed

ckwong99

Member
2005-05-24 6:41 am
HK
Hello all,

When I have do my A75's front end voltage gain adjustment, found the circuit not very stable. During adjust P1 & P2 VR acknowledge from OCR both input and output signal turn very bad. I think it may due to the circuit self-oscillate. Not sure to change the series resistor value of the Gate of FET would be helpful on it .

Every times I look my PSU display 30mA per channel using. The input signal shaping is good but no output voltage gain. There should not enough power for the circuit, then I try to adjust P1 & P2 VR . Somehow it suddenly surge more current about 70mA per channel then my both input and output signals collapsed.

1) Where and what resistors value can be modified ?

2) Can any other way to restrict the self-oscillation ?

3) Once change the values will the gain also be changed ?

4) Will it be affect by without matching the input differentiate pairs ?

Thanks for your attention.

CK
 
CK -

Did you use a local bypass on the power supplies? If you use the audioexpress boards (published layout) and don't add some capacitance to ground at the rail connection on the boards, they have a tendency to oscillate. try 100 uf with a 100 nf film on each rail.

Did you ever have any voltage gain? If always no gain, check that the feedback 75K resistor is in the right place and you don't have a jumper where it should be. (Are you trying to use 100% feedback from the output stage?) I made this mistake and it caused a channel to self destruct.

Not using matched devices in the differential pair makes it harder to balance the offset.

Sounds like you may have blown some FETs - the safe thing is to replace all. I used IRF610/9610 for the differential pair after blowing the last IRFD210/9210s that I had. Just bend the leads making sure they don't touch. I don't think it made a noticeable difference in the sound.

good luck
 

ckwong99

Member
2005-05-24 6:41 am
HK
Hello Bob,

A general PSU bought from the shop not a PCB type as you said. Shall I add the bypass caps on the rail ?

I counter checked with the gain circuit is correct. There the problem is not no gain that was only due to the starting current too small ( I had checked reference voltage on R9=4V and R7=0.4V). However, if increase the current adjusts P2 then oscillation occurred, although the voltage gained output shown but not high enough and worst.
I use R81=600, R3=15K, R27=open, R2=470, R29=open. I have chosen a partially feedback.

Recall that you taught me use IRF610/9610 instead of IRF210/9210. So they have already done. (Do you still remember our chat at this column about a month ago ? I also query about the bias FET I adopted K216. The final result is acceptable bias voltage can be under controlled by the resistors network no need make modify. Thanks for your advice.)

I read from Nelson’s DIY paper again find his comment when oscillation encounter can first increase the Gate resistor value. After change those value to 300 ohm (without change 4 input FET resisters). Seems the problem was cleared. Nevertheless its output waveform is not very sharp, clear and high enough.

Do you think the voltage gain will be decreased by using higher resistor value on the Gate from 100 to 300 ?

Thanks for your kinly help.

CK
 

Nelson Pass

The one and only
Paid Member
2001-03-29 12:38 am
ckwong99 said:
Do you think the voltage gain will be decreased by using higher resistor value on the Gate from 100 to 300 ?

On these parts, 300 ohms will have minimal effect on performance.
These values will affect the bandwidth of the output transistors,
but not 610's and 9610's.

Sometimes you just have to play around with the values a bit,
and then suddenly everything starts to work. Remember that
the front end board by itself is no capable of driving the output
RC network of the amplifier, but it is often included on the front
end board. If your front end appears to work at very low
frequencies, but not in the midband, then this would be your
problem.

:cool:
 

ckwong99

Member
2005-05-24 6:41 am
HK
Thanks to Nelson.

As you mentioned just play around a little on the Gate resistor value I had completely solve the problem. Do you consent to me to change 4 input FETs gate resistor value ?

The front end now have a gain in full range but not high enough and used to its output waveform not in pretty.

Up to this time I only test the front end separate from the output stage of the AMP. When unstable case occurred I found it impossible generate the output, as well the input signal also be collapsed.

CK
 

ckwong99

Member
2005-05-24 6:41 am
HK
Hello Bob & Nelson & All,

The last situation of my A75 front end is stable with a clean and clear voltage output operated in 100K Hz or above without any abnormal appeared. The following some changes were be done.

1) R17-R20 = 300
2) R21-R22 = 50
3) R27 = empty
4) R3 = 50K
5) R81 = 680
6) Q1-2 = IRF610
7) Q4-5 = IRF9610

However, its voltage gain is still not enough, the Vp-p OUT is about 20 times of Vp-p IN (1v). I found that it was clipped the peak when I adjusted higher the P1 & P2.

I used OCR measure output span of ground to R81 (towards Z3 & Z4) saw the output swing 10v positive and negative at about 0v. Nevertheless, I switch OCR input channel to "DC" then have same output but shifted upward 28V DC.

Can anyone help me to solve the gain problem ? If you need a further info or data just let me know.

Thanks for your kindly help.

CK
 
CK,

If I uynderstand you correctly, you have a voltage gain of 20 (27db), which is normal for this amp. Less than 2 V p-p (most CD players will output this much) will drive it to full output. If need more gain use a preamp.

What is odd is that with 50K for R3, no R27 and you did not indicate any changes to R2 or R29, your gain should be much higher. Did you make any changes not noted? With the gain you see, R3 should be around 15K as you noted in post 3

To answer an earlier question, yes, add bypass caps on the rails of each board. This should help the stability.

Note that R21 and R22 should be >75 ohms for stability (page 7 of part 1)

I'm not sure what you mean by OCR, is that a scope? Why are you measuring the voltage across R81? Remember that this amp is DC coupled and will amplify any DC on the input. If switching to DC results in a dc offset on the probes, you are putting DC into the negative input. Have you measured the DC offset at the output terminal?
 

ckwong99

Member
2005-05-24 6:41 am
HK
Hello Bob,

It do for my power amp so I prefer to drive it with per-amp.
The V gain about 25-26dB I measured but why the DC supply +/- 50V how I just have this +/- 10v signal outcome but not a +/- 30v signal. Recall that my output stage intend to apply +/- 60VDC, from this point of view, thus power from output stage would never boosted to the maximum by the front end . Do you agree on it ? ( Even I try the other ways increase the gain but end up clipping on the peak would happen.)

R29 replaced by 2k value.
R3 is a 50K VR, I saw it marked on about 13K value. Sorry for my mistake.
OCR I meant for oscilloscope you guess it right. I will not measure voltage between R81. I had just measured output from one end of R81 (which head on Z3&Z4) to GND. (I doesn’t make connection to ground with R31) Is it correct ? Because I verfied the waveform distorted seriously by R31 connected to GND. ( It much thanks to Nelson to give me a clue on pervious post.)

Normally the output can be displayed swinging at about the zero volt axis when “AC” to be selected on OCR. ( The signal was going through implicit capacitor before entry the OCR.) Not sure why I select to “DC” then the whole signal moved upward near to 28VDC.

By the way, I tried to match the differential pairs. Doing so I really to see the output waveform positive / negative cycle looks like to their opposite.

Thanks for you patience.
CK
 
At 26 dB gain, you should have no problem reaching maximum output swing. 2 volts on the input will give you 40V out. Perhaps your source is a high impedance and cannot drive the amp to maximum output. Are you using single ended or balanced inputs? What voltage can your source provide unloaded?

Increasing R3 will increase gain, at the expense of increased distortion because less feedback being applied. Whether it is audible or not remains to be seen. Closed loop gain should be set by the voltage divider formed by R3 and the parallel combination of R2 and R29. Decreasing R29 to the 1.5K specified will also increase gain.

Once you determine the value you want for R3, I suggest that you replace the pot with a fixed resistor. Feedback resistors should be best quality - if you're going to only have one "audiophile grade" resistor, R3 is where to use it.

I presume that you set the offset using a voltmeter and the DC voltage at R81 referenced to ground is <50mV. If this actually changes when you go from AC to DC on your scope, it is because your scope has DC on its input. Is the offset measured with a voltmeter or looking at the scope trace? Check with a VM. I have seen a jump on my scope when changing that settles out in a minute or two. Must be a capacitor charging or discharging somewhere.
 

ckwong99

Member
2005-05-24 6:41 am
HK
I have used a mini-signal generator ( A kit bought from market ). Its output direct connect to the input of the circuit. I think the output level would not high enough but I don't know how the impedance of it.

Next, I will try use it input to my preamp first and then take output to the circuit.

I knew the equation of gain, where pressed on the A75's DIY . Somehow the gain it never conform the equation. I can't explain and understand it.

Sure that I will replace some majority components.

The offset measured from the OCR not the DMM. First step to see what the quality and gain of output yield. Second, to balance the DC offset by using DMM. I think DC offset is little bit because the signal centered at about the zero volt axis.

Does it have too large DC level (measured from OCR use "DC") that was be riding on the AC signal according to my experience ? Is it possible ? Or it was abnormal according to your recognize ? Somewhere had capacitor to be charged after I do measure.

CK
 
Maybe you can buy/borrow a voltmeter - It is possible that there is that much offset, but if everything else seems to be right try measuring witha different tool.

Also, check for DC on the input - your signal generator may have some DC that is getting amplified.

Try measuring offset with the + input shorted to ground.

Is the inverting (-) input shorted to ground if you are using a single ended source?

Do you have R30 in place? What value is it? check for cold solder joints here.
 

Nelson Pass

The one and only
Paid Member
2001-03-29 12:38 am
In case you were not aware -

In the PC artwork of the original A75 boards
in Audio Amateur, there are three resistors
mis-labeled. The schematic is correct.

R22 on the PCB is R24 on the schematic


R26 on the PCB is R81 on the schematic


R24 on the PCB is R26 on the schematic

In addition to this, it is worth noting that
a modification was made in improve the reliability
of the front end when the amp is overdriven. Two
Zener diodes are added as seen in the diagram:
 

Attachments

  • a75mod.gif
    a75mod.gif
    5.4 KB · Views: 343

ckwong99

Member
2005-05-24 6:41 am
HK
Thanks Nelson.

I just only follow the document from DIYPass provided. Not sure there will have same problem to Audio Amateur.
Indeed, I found the mistake label on PCB, but you must aware it when stuffing components. Anyway I will check it again with respect to your reminding.

I think my circuit still have some problem. Perhaps due to above wrong things was be ignored. An unusual matter report, I touch the FETs feel their temperature very high. ( about 45-50 degree). Especially which happen in Q3,Q6,Q9,Q10 the folded cascode part.

Add two zeners into the circuit is wroth to do.

Wait for my acknowledgement later , as hurry as I test it again on this week end.

CK
 

ckwong99

Member
2005-05-24 6:41 am
HK
Hello there ,

Especially for BobEllis & Neslon :>

My A75 found an important mistake but actually I can't ensure this is a crucial to affect the output signal... (Now the output gain is satisfy but still the waveform is not much pretty and with phantom.)

I am inattention use the IRF 9610 instead of IRF 9510. By datasheet indicate that IRF 9610's RDS on is 3 ohm but 9510 is 1.2 ohm. Is there the reason that I never get satified output ?

CK
 
CK - I doubt that using IRF9610s is the cause of your dissatisfaction. RDS is pretty irrelevant for this application. More important is Ciss, which is similar for both devices. The 9610 is a little lower, which should extend the top end a hair.

It sounds like you may still have oscillation issues. Have you increased R21 and R22 to >75 ohms?

Are you testing a completed amp or just the front end? remember that the front end cannot drive the zoebel by itself. If the output stage is underbiased, you will get lots of crossover distortion. be sure you have at least 100mA per device. Wait for things to warm up a bit - my A75s take a couple minutes for the bias to come up.
 
ckwong99,


>> I am inattention use the IRF 9610 instead of IRF 9510. By datasheet indicate
>> that IRF 9610's RDS on is 3 ohm but 9510 is 1.2 ohm. Is there the reason
>> that I never get satified output ?

NO, mine is assembled with IRF 9610 and is working perfectly ; follow the hint's of Nelson and after add 4 * 100 microF on the Pcb, one for each +- powersupply connector.

Good luck.

Alain.
 

ckwong99

Member
2005-05-24 6:41 am
HK
Hello Bob & Alain,

I only test the single channel of front-end and without other connection from its output. R21 & R22 now is using 100 ohm. Should these value need to increase a bit more ?

From your experiences 9610 not be a main harm, so only left the event is regarding the circuit stability. My temporary DC supply had already installed 100uF + .1uF capacitors on each + / - rails.

My PCB C9 and C10 are empty. Does it require to put it back ?

Thanks all of you.
CK
 
ckwong99,

From A75 part1:
Two new capacitors, C9 and C10, add some lag compensation in the
event that a particular choice of components and layout are not
adequately stable. We offer no values, since we did not use them in the
amplifiers we built, nor have we encountered the problem in a range of
prototypes. If necessary, your choice of capacitors will be for whatever
prevents high frequency oscillation, and may range from 5 to 39pF. If
oscillation occurs, consider raising the values of the gate resistors on the
MOSFETs first.

I didn't use any... but you could try 10 pf to start.

>> My temporary DC supply had already installed 100uF + .1uF capacitors on each + / - rails.

--> These MUST be on the A75 front end PCB !!! just near the -+ connections to the board.
The problems comes from the wiring, between power supply and front end Pcb.

Alain.
 
Just to emphasize the point made by Alain, if your 100 uf and 100nf are not right at the power terminals of the board, you are very likely to have oscillation. you'll want a set on each power connection, so 4 of each value are required.

I didn't have a scope when I built my first A75, so I used 10pf micas for C9 and C10 because I found someone else saying that worked without hurting the sound. After blowing up a few sets of outputs on the first, I decided to play it safe on my second A75, and used them again.

If you don't have caps handy, a couple inches of hookup wire tightly twisted together should give you ~5pf to see if that helps. NP mentioned this a couple years ago, and even provided a picture.