Wendy200 a gainclone based on the BPA200

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello all,

This is my first attempt to design and realize a BPA200 based amplifier with 200W per channel.

I have named it Wendy200 after my better half of course !

The amp features:

- 2x LM4780 power amps per channel, in an BPA200 arrangement as seen in figure 17 of the AN-1192 application notes (using the non-inverting part of the circuit only)

- LM4562 instrumentation amplifier front-end / buffer

- THS4131 for voltage gain

- 0.1% quality resistors all around

- SMD components all around for compact size and short signal path

The front end and power supply scheme is quite similar to the Sympatico amp from my favourite twistedpearaudio.

The attachment shows the schematic.

The instrumentation amp has a gain of 2, the THS4131 has a gain of 1.1 and the power amps have a gain of 20+20=40 (due to bridged configuration)

I just want to share it with you all to debug, improve it and choose the right components.

Thanks
Evan
 

Attachments

  • wendy200-v2.png
    wendy200-v2.png
    85.9 KB · Views: 2,353
Alternative design

Alternatively this design (attached) could be followed, which theoretically is the same as the above (I think), the only difference is the negative feedback point of the THS4131 is now from its output directly (as in the THS4131 datasheet applications) and not the output of the LM4780. Is that a better implementation and if so why ?

Another question I have is what type of capacitor should I use for C1 - C4 ? In the national's application notes they use both electrolytic and non-polarized ones - which is best? Would ceramic or niobium oxide ones be ok? Another option is to completely ditch them, is that going to be okay?

Also is 3 watts rating enough for R13 - 16 ? It's difficult to find 5W resistors.

Cheers
Evan
 

Attachments

  • wendy200b.png
    wendy200b.png
    35.4 KB · Views: 1,960
All depends on what your looking for.
My personal tastes would be to take the output from the LM4870.
One suggestion increase C1-C4 from 22uf to 47uf, I tried going to 100uf and low frequency motor control became an issue 22uf is to low for solid low frequency's 47uf is the best compromise.
3 watt output resistors are fine, I have used 3 watt ones in all my BPA's with no issue.
 
tiltedhalo said:
All depends on what your looking for.
My personal tastes would be to take the output from the LM4870.
One suggestion increase C1-C4 from 22uf to 47uf, I tried going to 100uf and low frequency motor control became an issue 22uf is to low for solid low frequency's 47uf is the best compromise.
3 watt output resistors are fine, I have used 3 watt ones in all my BPA's with no issue.


Thanks for your advice tiltedhalo

My aim is to achieve good low frequency response so I will follow your advice to use 47uf for C1 - C4

What material did you use for these capacitors?

Evan
 
tiltedhalo said:
OK hold on to your hat! I use 47uf 24V tantalum's jumped with a .47uf metal film WIMA. I tried black gates, muse, oscon's jumped with poly and not jumped and found the best sound, response, and control from the tants and wima combo.
here is a pick of my daily user ...........


Wow nice project tiltedhalo! Seems like you've done a thorough job.

As for the cap material ... I see !! I will try different ones and pick the best then.

One more question, when you say you are using 2 caps "jumped", do you mean in series or in parallel?

In that case you would get 94uf or 22uf respectively and not the 47uf you suggested right?

Cheers
 
A 47uf paralleled with a .47uf metal film WIMA.
I have built a bunch a BPA's to come up with what to my ears and test equipment sounds the best.
IMO I think the combo of the of the THS4131 and LM4562 will be very clinical to bright sounding, but that's all personal taste.
I will never listen to anything else, I first thought the idea of chip amps stupid, then with all the hype I built one and have been hooked from then on.
I replaced a Carver 4.0t and Krell KSA250 for my BPA300 and 200.
 
tiltedhalo said:
A 47uf paralleled with a .47uf metal film WIMA.
I have built a bunch a BPA's to come up with what to my ears and test equipment sounds the best.
IMO I think the combo of the of the THS4131 and LM4562 will be very clinical to bright sounding, but that's all personal taste.
I will never listen to anything else, I first thought the idea of chip amps stupid, then with all the hype I built one and have been hooked from then on.
I replaced a Carver 4.0t and Krell KSA250 for my BPA300 and 200.


Thanks for your reply once again

I get it now 47uf + 0.47uf

As for the op-amps yes I think you are right but I prefer a power-amp to be clinical or as transparent as possible.

I can then inject some warmth by using a valve pre-amp.

I think BPA200 with LM4780's is perfect because it's compact and paths are short.

In BPA300 the chips are getting too far away from each other plus it's getting expensive or at least of a similar price tag as the class D amps.

Cheers
 
I would use each dual chip to amplify one inverted and one non inverted channel rather than each chip amplifying the same polarity channels.

I agree, C1 to 4 need to be larger.
I would try 100uF or 150uF and try to solve the PSU motorboating problem rather than restrict the LF content.
Each chipamp is driving the equivalent of the load impedance. If your load is 8ohms then each individual amplifier needs +-20mF for C1=150uF. But the inverted signal passing through the dual chip may allow this smoothing capacitance requirement to be relaxed substantially.
A dual chip passing the same signal would require the +-20mF for each half equaling +-40mF per dual chip and +-80mF for the complete amplifier.

Have a read of W.Jung's paper on composite opamps.
Most of what he has to say applies to your composite.
The gain in the final half is determined by the overall gain you require. I think he suggests something like 2 times. If I recall that gain advice correctly then the gain stability of the power chipamp will not allow you to follow Walt's composite model. i.e. it might not work.

Also read Walt on balanced audio.
He advises very high impedance on the balanced inputs. i.e. replace 100k with 1M5.
 
AndrewT thanks for your educated reply.

Note that I am not any expert in design so your suggestions have triggered a lot of questions to me:

AndrewT said:
I would use each dual chip to amplify one inverted and one non inverted channel rather than each chip amplifying the same polarity channels.

1. What would be the benefit of that ? Balancing the thermal load maybe? I would prefer the poweramps to drive the same polarity for a simple PCB.


I agree, C1 to 4 need to be larger.
I would try 100uF or 150uF and try to solve the PSU motorboating problem rather than restrict the LF content.
Each chipamp is driving the equivalent of the load impedance. If your load is 8ohms then each individual amplifier needs +-20mF for C1=150uF. But the inverted signal passing through the dual chip may allow this smoothing capacitance requirement to be relaxed substantially.
A dual chip passing the same signal would require the +-20mF for each half equaling +-40mF per dual chip and +-80mF for the complete amplifier.

2. I think understand your point. So presumably "motorboating" is caused by RF inteference in the negative feedback points, right? If so, how can this interference be rejected? Use of filter caps elsewhere maybe?

Have a read of W.Jung's paper on composite opamps.
Most of what he has to say applies to your composite.
The gain in the final half is determined by the overall gain you require. I think he suggests something like 2 times. If I recall that gain advice correctly then the gain stability of the power chipamp will not allow you to follow Walt's composite model. i.e. it might not work.

Also read Walt on balanced audio.
He advises very high impedance on the balanced inputs. i.e. replace 100k with 1M5.

3. I would really appreciate if you could point me to these 2 articles (links or attachments?). I have only read the really basic stuff in op-amps.

Ideally I want to keep the gain of the composite front-end to a minimum and really use the op-amps as buffers and for noise rejection (oh and XLR inputs too).

So, If I set the instrumentation amp gain to 1 (by removing R21) and change R22 & R23 from 22k to 44k to get a gain of 2, would that be in accordance to W. Jung's recommendations if I understand well?

Thanks again and apologies for the many questions... I am over excited about this project !

Evan
 
chatziva said:
1. What would be the benefit of that ? Balancing the thermal load maybe? I would prefer the poweramps to drive the same polarity for a simple PCB.
no.
it is to reduce the transient current demand in the power supply pins.
Rather than draw double the current through each pin alternately, each pin draws peak current simultaneously but at sufficient to to supply the load (not doubled).


chatziva said:
2. I think understand your point. So presumably "motorboating" is caused by RF inteference in the negative feedback points, right? If so, how can this interference be rejected? Use of filter caps elsewhere maybe?
motorboating is a low frequency instability, more common to valve (tube) circuits.
Arranging the RC time constants of SS circuits almost always eliminates it from our designs, but

I tried going to 100uf and low frequency motor control became an issue 22uf is to low for solid low frequency's 47uf is the best compromise.
implies that motorboating can still happen when the time constants are selected arbitrarily.
The problem is almost certainly down to the PSU setting the LF filter rolloff.
When the input filter (which your circuit does not have) should set the bandwidth.
 
AndrewT said:
...implies that motorboating can still happen when the time constants are selected arbitrarily.
The problem is almost certainly down to the PSU setting the LF filter rolloff.
When the input filter (which your circuit does not have) should set the bandwidth.

I think I will try 100uf and will include an input filter if motorboating occurs :smash:

When you say "input filter" do you mean a "signal-in" filter network?

What would be a good one (values) assuming motorboating occurs?

Thanks
Evan
 
AndrewT said:
no.
it is to reduce the transient current demand in the power supply pins.
Rather than draw double the current through each pin alternately, each pin draws peak current simultaneously but at sufficient to to supply the load (not doubled).


I plan to use common V+ and V- for both power amps so the arrangement you are suggesting is not making any difference unless I use separate supplies for each power amp, right?

Evan
 
it makes a difference to the current flowing to the power pins and through the power pins.
This in turn will affect the way the PSU interacts with the chipamp.
That is why I suggested that the PSU smoothing requirement may be relaxed if you use each chipamp to amplify both polarities.
 
chatziva said:
I will try 100uf and will include an input filter if motorboating occurs ...........When you say "input filter" do you mean a "signal-in" filter network?

What would be a good one (values) assuming motorboating occurs?
The input filter should be about an octave higher than the filter effect of the PSU and about half an octave higher than the NFB filter.

I use ~90mS @ input, 130mS to 150mS @ NFB and 160mS to 200mS @ PSU.
90mS~=F-3db@1.8Hz.
 
AndrewT said:
The input filter should be about an octave higher than the filter effect of the PSU and about half an octave higher than the NFB filter.

I use ~90mS @ input, 130mS to 150mS @ NFB and 160mS to 200mS @ PSU.
90mS~=F-3db@1.8Hz.


Maybe a drawing / calculations of RC values would really help me understand this.

What is NFB ?

So does this mean that you recommend the use of input cap ? (like the one shown in the national's application notes i.e. 0.47uf + 47k ).

I was trying to avoid having an input cap... (is that good practice?)
 
AndrewT said:
it makes a difference to the current flowing to the power pins and through the power pins.
This in turn will affect the way the PSU interacts with the chipamp.
That is why I suggested that the PSU smoothing requirement may be relaxed if you use each chipamp to amplify both polarities.


Sorry I don't understand this.

In either arrangement all the V+ pins will be connected to a common line for all 4 power-amps and the same goes for V- pins.

So the transient currents in the possitive and negative rails should be the same in any of the 2 arrangements, no?

I plan to use 4x 10,000uf per channel for smoothing, the schematic will follow soon...
 
I believe the point Andrew T. is trying to make with one half of each chip inverted and non inverted is to reduce the current demand in half at the chip, reducing the loading effects on the PS, chip pins and PCB traces.
When one side of the chip is drawing negative current the other is drawing positive, instead of positive positive negative negative, this will help with the need of heavy decoupling at the chip, and the LM4870 has small pins and multiple power pins, and small traces to connect these it can introduce capacitance and resistance variables at the chip.
Andrew T. correct me if I am wrong.
As for motor boating I have never had an actual problem with this, I have had issues with capacitive saturation at heavy signal swing with larger caps.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.