Share |

Dynahi: The AMB way

Kevin Gilmore's higher-powered version of the dynalo

Dynahi: The AMB way

Postby mattcalf » June 12th, 2010, 7:33 am

Ti has recently taken stock of the Kevin Gilmore Dynalo/hi series and as a future Dynahi builder I have been doing a bit of research. Which led me to stumble across Ti's altered schematic for the Dynahi and figured I'd post it and the justifications that he gives for it (that I can find). Also, this thread can hopefully serve as a general thread for any future builders. I know I'll have a few queries!

For Ti's original build, photos, measurements and specs go here.
His schematic can be accessed here.

The are a number of modifications to the original schematic:
  • Parallel a 680R resistor and 10k Trimpots to adjust bias and offset, you should pre-set the trimpots such that you measure 500R across them after installing it on the board, along with the 620R or 680R resistor. This would then be the initial power-up setting. (setup procedure using them here)
  • A 475 ohm resistor in series with the wiper to the amp board to buffer the board a bit from being shorted completely to ground (volume pot minimum condition) or connected directly to the output stage of the source (volume pot maximum condition).
  • Optional (well it's all optional I guess, but this one especially?): Reducing the gain to 7 by changing the global feedback resistors to 1.21KΩ and 200Ω (from 2KΩ and 100Ω) and increasing the feedback compensation cap value from the suggested 33pF.
  • Substitute the OP27 for a OPA134PA as it offers much lower DC offset.
  • Kevin's schematic showed no input resistor to ground, but the board is laid out for a 10K resistor. If you have a <100K volume pot wired up at the input, then 10K is too low a value. Anywhere from 10K to 1M is more "the norm" for a JFET input amp.

Please post anything I've missed and I'll add it in. I realise this is all copy and past, so a bit redundant, but makes it a bit easier for any future builders (I hope).
amb wrote: Like a pure class A amp should, the maximum heat dissipation occurs when the amp is idle. It will actually run cooler when playing loud.
mattcalf
 
Posts: 58
Joined: April 4th, 2010, 8:29 pm
Location: Far North Queensland, Australia
Country: Australia (au)

Re: Dynahi: The AMB way

Postby mattcalf » June 12th, 2010, 7:33 am

A personal question now:
In a Dynamite setting, what VA rating on a transformer would I need for dual Sigma22s (heatsinked by conrad) to comfortably drive my speakers (a 6 ohm multiway speaker and has dips down to 4.5 ohm and is less than 5 ohm from 100Hz to 600Hz)?

Thanks again,
Matt.
amb wrote: Like a pure class A amp should, the maximum heat dissipation occurs when the amp is idle. It will actually run cooler when playing loud.
mattcalf
 
Posts: 58
Joined: April 4th, 2010, 8:29 pm
Location: Far North Queensland, Australia
Country: Australia (au)

Re: Dynahi: The AMB way

Postby amb » June 12th, 2010, 1:16 pm

mattcalf wrote:[*]Optional (well it's all optional I guess, but this one especially?): Reducing the gain to 7 by changing the global feedback resistors to 1.21KΩ and 200Ω (from 2KΩ and 100Ω) and increasing the feedback compensation cap value from the suggested 33pF.

With the lower gain I observed a peaking of the frequency response in the ultrasonic range, which means that the amp is borderline unstable. For those who don't have a wideband function generator, an oscilloscope and a wide selection of compensation cap values to tune the amp with, I recommend staying with the default high gain. This of course means that the dynahi would be suitable for high-impedance headphones only.

In a Dynamite setting, what VA rating on a transformer would I need for dual Sigma22s (heatsinked by conrad) to comfortably drive my speakers (a 6 ohm multiway speaker and has dips down to 4.5 ohm and is less than 5 ohm from 100Hz to 600Hz)?

While the dynahi has enough voltage swing and could be equipped with large heatsinks, I don't really recommend it as a speaker power amp at all. The output stage is four paralleled pairs of BJTs, but those output devices are rated only 1A max each, for a total theoretical max current of 4A. They are driven directly from the VAS (voltage amplification stage). Since BJTs are current-current amplifiers, the VAS has to supply the output current of the amp, divided by the Hfe of the output stage BJTs.

Assuming an 8 ohm load with a maximum output voltage swing of 21V peak, that means 2.6A of peak output current. This is 65% of the total rating of the output devices. Also each output branch has 10 ohm emitter resistors (two 20 ohms in parallel, and there are four such pairs, so the equivalent is 2.5 ohms), and they will severely reduce the output swing (at 2.6A there will be 6.5V drop on each resistor). This is not to mention that at such currents each resistor would have to dissipate over 2W of power, but the specified RN60D resistors are only rated at 1/4W, and could realistically be called 0.5W resistors.

OK, so let's say if we used much lower value output resistors (e.g., 0.47 ohm) to get around the voltage drop issue (which, in reality could lead to BJT thermal runaway but let's ignore that for now), and assume the output transistors' Hfe is 150, then its base drive must be 2.6A / 150 = 17mA. But the total current flow through the VAS is only around 30mA or so, so more than half of that current will have to go into driving the output BJTs, clearly not ideal for high performance.

If your speakers have lower impedance than 8 ohms, then the situation becomes worse. Especially in a "dynamight" configuration where the output voltage is doubled, so the output current is doubled as well.

The dynahi was designed to be a headphone amp, not a power amp. It could be used to drive speakers as is, and it would probably even achieve reasonable volume, but as a speaker power amp, it needs larger output devices, an addition of a pre-driver stage before the output stage to offload the VAS, reduced output emitter resistor values and a proper output stage biasing circuit with thermal-feedback.
User avatar
amb
Site Admin
 
Posts: 6639
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: Dynahi: The AMB way

Postby mattcalf » June 12th, 2010, 6:26 pm

Thanks for the replies AMB. Was reading about it last night and this thread got me thinking that it might be capable both ways, but you're explanation is the advice I'll take. :)

Not to derail my own thread, but how much better would a balanced B22 handle both duties?
amb wrote: Like a pure class A amp should, the maximum heat dissipation occurs when the amp is idle. It will actually run cooler when playing loud.
mattcalf
 
Posts: 58
Joined: April 4th, 2010, 8:29 pm
Location: Far North Queensland, Australia
Country: Australia (au)

Re: Dynahi: The AMB way

Postby amb » June 12th, 2010, 8:42 pm

mattcalf wrote:Thanks for the replies AMB. Was reading about it last night and this thread got me thinking that it might be capable both ways, but you're explanation is the advice I'll take. :)

Yup, note that rsaavedra could only describe things in anecdotal ways, with no hard numbers (as I said, "would probably even achieve reasonable volume"). But he is quite mistaken to describe it as "pure class A sound in speakers". When the dynahi is pushing that much output current, it no longer operates in class A.

Not to derail my own thread, but how much better would a balanced B22 handle both duties?

β22 has 17A-rated output MOSFETs. The current rating is 4x+ those of the dynahi's output transistors combined, but β22 is thermally-limited due to those MOSFETs' TO-220 package. There is only so much heat that could be transferred on a modest device size, even when a large heatsink is used.

β22's 0.47 ohm output resistors will not cause big voltage drops like the 10 ohm ones in the dynahi would. MOSFETs have negative temperature coefficient, so they won't run away like the BJTs. MOSFETs are voltage-current devices, so they require almost no current drive from the VAS (except to charge the gate capacitance on fast transients, but β22's dynamic cascode topology virtually removes the gate charge requirement.

So, to make a long story short, β22 was also designed to be a headphone amp and is best used as one, but it has much less fundamental limitations as the dynahi as a speaker amp. The only real limiting factor is heat, and overheating the MOSFETs usually results in rather unfortunate damage. This is why there is a β24. ;)
User avatar
amb
Site Admin
 
Posts: 6639
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: Dynahi: The AMB way

Postby mattcalf » June 12th, 2010, 9:09 pm

Okay thanks for that, asking for a bit much for a world class headphone amp that can man-handle my speakers. There's always trade-offs, I guess.
I'll have to think about a separate solution for each then.

amb wrote:This is why there is a β24. ;)

I wouldn't trust myself building a B24, not for a while anyway. :)
amb wrote: Like a pure class A amp should, the maximum heat dissipation occurs when the amp is idle. It will actually run cooler when playing loud.
mattcalf
 
Posts: 58
Joined: April 4th, 2010, 8:29 pm
Location: Far North Queensland, Australia
Country: Australia (au)

Re: Dynahi: The AMB way

Postby Nebby » June 12th, 2010, 9:54 pm

The β24 is a BEAST. Not to mention a beast to move....I personally came to a very good understanding as to why Amb chose to put handles on his.
Nebby
 
Posts: 316
Joined: April 23rd, 2010, 4:08 am
Location: Maryland
Country: United States (us)

Re: Dynahi: The AMB way

Postby mattcalf » June 13th, 2010, 8:35 pm

Gday Ti,
Just curious on the dimensions of the dynahi board. From what I can find, the last group buy had the dimensions at 4"x4.95"x0.062". Is that correct?

Thanks,
Matt.
amb wrote: Like a pure class A amp should, the maximum heat dissipation occurs when the amp is idle. It will actually run cooler when playing loud.
mattcalf
 
Posts: 58
Joined: April 4th, 2010, 8:29 pm
Location: Far North Queensland, Australia
Country: Australia (au)

Re: Dynahi: The AMB way

Postby amb » June 13th, 2010, 9:19 pm

mattcalf wrote:Just curious on the dimensions of the dynahi board. From what I can find, the last group buy had the dimensions at 4"x4.95"x0.062". Is that correct?

Yes.
User avatar
amb
Site Admin
 
Posts: 6639
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)


Return to dynahi headphone amplifier

Who is online

Users browsing this forum: No registered users and 0 guests

Quick Links

AMB audio DIY resources
AMB audio shop