A variation on this would be - which type or general topology of speaker/passive XO is the hardest on the amp and causes distortion to be increased.
THx-RNMarsh
RNMarsh (4339)
Several models of speakers exist which can be used...
Is there a universal OPS that keeps distortion at its lowest with any and all speaker and XO types? Or are there two... one for electrostatics and one for dynamic loads? And, are the SOA protection different?
I absolutely agree with you. When designing the amplifier is very important to choose the output stage (OPS). Today there are many circuits from the output stages from simple " twos" ... "Fours" Darlington , and to more complex circuits using transistors Shiklai , " diamond " transistor circuit Briston and more. And that's not counting the classes B, AB, A, non-switching and ather.
I know of loads models do not allow the load to bring the amplifier to the real work of the amplifier with speaker system . The greatest idea of the quality of OPS provides a simple square wave generator frequency of 2 kHz and an amplitude of + -5 V is connected in series with a load of 4 or 8 ohms. Quality measure is the error signal = Uin - Uout
Best regards
Petr
Several models of speakers exist which can be used...
Is there a universal OPS that keeps distortion at its lowest with any and all speaker and XO types? Or are there two... one for electrostatics and one for dynamic loads? And, are the SOA protection different?
I absolutely agree with you. When designing the amplifier is very important to choose the output stage (OPS). Today there are many circuits from the output stages from simple " twos" ... "Fours" Darlington , and to more complex circuits using transistors Shiklai , " diamond " transistor circuit Briston and more. And that's not counting the classes B, AB, A, non-switching and ather.
I know of loads models do not allow the load to bring the amplifier to the real work of the amplifier with speaker system . The greatest idea of the quality of OPS provides a simple square wave generator frequency of 2 kHz and an amplitude of + -5 V is connected in series with a load of 4 or 8 ohms. Quality measure is the error signal = Uin - Uout
Best regards
Petr
I still have some questions -
As DIY'ers here, which OPS topology gives the most bang for the buck in regards to load/distortion? I mean without complex error correction, too. The simplest OPS with the lowest cost that has the best distortion results? OPS topology only.
Is a 2EF better than a Sziklai to use on a simple CFA (or VFA) circuit? Or is it just a pair of FETs for the OPS? Simple, low cost vs distortion under real load conditions?
-RM
As DIY'ers here, which OPS topology gives the most bang for the buck in regards to load/distortion? I mean without complex error correction, too. The simplest OPS with the lowest cost that has the best distortion results? OPS topology only.
Is a 2EF better than a Sziklai to use on a simple CFA (or VFA) circuit? Or is it just a pair of FETs for the OPS? Simple, low cost vs distortion under real load conditions?
-RM
Last edited:
IMO, the Bartholomew Nicholas Locanthi II T-circuit, equipped with modern high Ft RET bipolars. With the usual high Ft devices (in)stability countermeasures and optimally biased.
Mosfets do allow a higher ULGF (and hence more feedback to linearize) however, intrinsically, the distortions of a mosfet OPS are larger compared to power bipolars. Mosfet or bipolars, pick your poison, Mr. Cordell likes mosfets, Mr. Self likes bipolars. This is another endless discussion.
Again IMO, a 2EF is recommended for low power (<100W) amplifiers only.
Sziklay OPS (also known as CFP OPS) are (again) IMO dangerous in terms of stability.
And very resilient , as well. .05% and will run 30 years non-stop.Cheap to
make ... another 1$ for predrivers.
Of all the EF3's I've repaired, only 1% have had hard faults. Dried out
thermal compound and caps only.
The drawbacks are that they are harder to thermally compensate and a slight loss of peak voltage swing.
OS
has anyone combined OPS using both BJT and MOSFETs? Say - use an EF off the gain stage and then to two seperate EF [with bias speaders] for the two different EF output transistor types?
Maybe with a LP XO for the BJT to only handle LF audio. Both outputs would be combined in parallel and sent to spkr.
Just wonder what that would sound and measure like.... high amperage from the bjt and hf operation with the fet.
The reason I bring this up is because i just replaced a VFA high current bjt OPS amp (one I blew up) with a CFA with fet OPS. The bass energy is clean and clear but lacking current for the 'weight' of the bass sound. [I have to drive 4- 15 inch Cerwin-Vega bass drivers]
Thx-RNMarsh
Maybe with a LP XO for the BJT to only handle LF audio. Both outputs would be combined in parallel and sent to spkr.
Just wonder what that would sound and measure like.... high amperage from the bjt and hf operation with the fet.
The reason I bring this up is because i just replaced a VFA high current bjt OPS amp (one I blew up) with a CFA with fet OPS. The bass energy is clean and clear but lacking current for the 'weight' of the bass sound. [I have to drive 4- 15 inch Cerwin-Vega bass drivers]
Thx-RNMarsh
Last edited:
There have been some designs on DIYA that have combined BJT/FET.
1.) BJT driver /vertical MOSFET OP(EF2).
The "quasi amp" has been built many times over. Able to produce a lot
of "grunt" (200-300W) from 2-3 pair vertical FET's.
The drawbacks are the loss of output swing , a Vbe is still needed , hexfet's
need to be biased pretty heavily. (more heat)
2.)BJT driver /lateral FET - easy to build , no tempco to speak of.
3.)The opposite - mosfet driver / BJT OP (EF2) .....
Have seen this , too ... actually , pretty cool
driver negates the OP'stempco slightly , harmonic structure is more H2 'ish.
Edit - must run a "hotter VAS" - gate capacitance ... can run a large OPS (no "droop")
4.) the "oddball" ... like the "Naska" amp . BJT drivers running an asymmetrical
BJT/latFET OP - "quasi style". More distortion with an almost valve like harmonic structure.
Interesting .....
OS
Last edited:
Well, it's unanimous... 3EF -- EF3 for >100W and EF2 for ,100W
However, I was refering to an OPS that used both fet and bjt power devices. not driver/ops. Two parallel OPS combined together at the speaker output using same front end.
In a similar thought - how do you synthesis a IGBT? Any really good coomplimentary IGBT?
Thx-RNMarsh
However, I was refering to an OPS that used both fet and bjt power devices. not driver/ops. Two parallel OPS combined together at the speaker output using same front end.
In a similar thought - how do you synthesis a IGBT? Any really good coomplimentary IGBT?
Thx-RNMarsh
Well, it's unanimous... 3EF -- EF3 for >100W and EF2 for <100W
However, I was refering to an OPS that used both fet and bjt power devices. not driver/ops. Two parallel OPS combined together at the speaker output using same front end.
In a similar thought - how do you synthesis a IGBT? Can we make such composite with off-the-shelf parts?
Or, do any really good complimentary IGBT exist for our CFA?
Thx-RNMarsh
However, I was refering to an OPS that used both fet and bjt power devices. not driver/ops. Two parallel OPS combined together at the speaker output using same front end.
In a similar thought - how do you synthesis a IGBT? Can we make such composite with off-the-shelf parts?
Or, do any really good complimentary IGBT exist for our CFA?
Thx-RNMarsh
Last edited:
Don't know. I have measured some of the Halo power amps from years ago (granted perhaps not his SOTA affort) and they have too much thd at higher audio freqs.
How would you make a very low distortion high power IGBT with discretes?
And, Which n/p compliments for driver fets? ... below and above 100W.
Thx-RNMarsh
How would you make a very low distortion high power IGBT with discretes?
And, Which n/p compliments for driver fets? ... below and above 100W.
Thx-RNMarsh
Last edited:
I have heard nothing good about IGBTs in audio output stages - especially latch up prone.
There was a flurry of interest in the 80s and early 90s and they then faded. They were designed for switching high voltages with low on switching losses - not really for linear applications. Great for motor control applications.
There was a flurry of interest in the 80s and early 90s and they then faded. They were designed for switching high voltages with low on switching losses - not really for linear applications. Great for motor control applications.
I've successfully worked on a design in simulation using Lateral FET's for normal demands and Vertical FET's running at zero standing bias for when the going gets tough. It was based on idea I put forward some years ago and relies on the Vgs properties of the different technologies.
These show the basic idea and the actual currents in each device in a CFP stage together with the load current. The lateral FET's max out at 6 amps (green) with the verticals taking over from that point as the loading becomes more severe (blue).
Attachments
Only when in Class A. In Class AB, around the crossover, mosfets are in the so called subthreshold conduction mode. They have exponential Id-Vgs there, by no means better than the bipolars Ic-Vbe.
It is hard to find suitable MOSFET for VAS duty (good combination of Ci and fT and may be RdsOn), especially a matched push-pull pair. I have used the ZVN part as in JLH amp and Fetzilla, IRF510(?) as in Ayre V3, but I have no access to the newer Toshiba hexfet. I think the ZVN is the best part available. There's a thread where I posted a list of MOSFET suitable for VAS duty.