F5 w/- 32v rail questions

Allan

Member
2004-03-24 6:38 am
Perth
Hi,

I'm upping the rails on my F5 to 32V (600VA 24v transfo) and have read the F5 Turbo manual where it says I've to also consider increasing R9 through to R12 to 5 watt, I intend changing these.

I'm a little confused though. If I only up the rails to 32V and change R9 through R12 to 5 watt, will this configuration alone still be capable of driving loads down to 2 ohm without burping?

Whats throwing me is I've read elsewhere that a stock F5 wont burp driving loads down to 2ohm, but in the F5 Turbo manual it goes onto asking "Want to drive 4 ohms with this? You can do it, but I recommend that you consider the upgrade to the F5 Turbo V1". However, the V1 also has the limiters removed and an additional set of output devices.

So where do I stand with loads down to 2 ohm if I only increase the rails to 32V and change R9 through R12 to 5 watt please?

If your wondering why I'm going with a "600VA".. I'm planning further upgrades later but for now I prefer to stage things up to be sure heat with my enclosure sinks doesn't catch me out... I wanna play with bias along the way

Many Thanks
 
Last edited:
Are you using the words from the F5 manual where it states that it will "drive a 2 ohm load without burping, and 1 ohm without misbehaving."

Your F5 is already stable down to 1 ohm loads. Is your goal more power into 2 ohms? Just for fun? For a specific speaker?

Either way, back to your original question...."will this configuration alone still be capable of driving loads down to 2 ohm without burping?"

Yes.

HOWEVER...

Increasing your rail voltages from the standard (~23V) on a standard F5 up to 32V will not give you more maximum power output into 2 ohms, PARTICULARLY if you lower the bias current in order to keep the heat on your sinks manageable per the article. That's actually going in the opposite direction from (what I think could be) your goal. See also the discussion (if you have not already) re: potentially cascoding your JFETs, particularly if you are using LS. Note, the cascode advice is not from my personal knowledge. I can't find the source, but I believe Nelson said the original Toshibas were borderline at 32V, but not a worry, but that the LS devices may break down earlier. Grain of salt unless I can find the original source. tl;dr - look it up or someone will likely chime in with the actual facts vs. my poor memory / conjecture.
 

Allan

Member
2004-03-24 6:38 am
Perth
thanks AIMH.

For now, yes my goal is more power simply by increasing the rail voltage to 32V. I'm aware of cascading but don't wish to take that route just now. I prefer to scale things up slowly, play with bias along the way and understand how things react first.. in particular the power & heat

I bought larger than needed sinks when I built my standard F5 so I could safely up the rails later.
Currently with 24v rails I run 0.61v bias and get ~48degC sink temps @ 22 -> 32degC room temp. Only once I've seen ~50degC on the sinks but never higher, that was a stinkin hot Assuie day.. I had the house open

So if I understand this correctly, unless I cascade to control heat thus up my bias, there'll be no power gain into loads down in the region of 4 ohm, seems its diminishing returns if I just increase increase the rails to 32v.



Thanks
 
Last edited:
Cascoding the input JFETs does not control the heat, nor will not increase your output power into any load. It's just to protect them from the higher rail voltages. Has nothing to relate to the output bias current.

tl;dr - if your goal is to increase available (max) power output into 2 ohms.... Personally, I'd start with upping your bias. Increasing the rail voltage without increasing the bias will not get your more power into 2 ohms. You'd just be adding more heat with no benefit. Again, I am sure someone will chime in if I'm way off base.

One of the most valuable things I've tried to do is lay out how much voltage and quiescent current are necessary for a power level into a load for SE and P/P Class A amps along with their associated idle dissipations => heat. Then, I see where the Class A envelope ends for the Class A/B along with max power (Voltage swing and current). I use some extremely basic assumptions that ZM helped me with. If you can do that, then you can see quickly whether increasing the rails / bias current / or both is needed to reach the power you desire into the load you need. With power generally quoted with an 8 ohm load, it's fascinating to see what tweaks will help 16ohm speakers or 2 ohm speakers. I'd try to lay it out for you for an F5, but I'm not quiiiiite there yet. Someone may be able to do it super-easy for you or help you with some hints / assumptions.
 
Last edited:

Allan

Member
2004-03-24 6:38 am
Perth
gotcha AIMH... so with 32v rails Jfet voltage rating is the real issue followed by bias => heat. OK then I'll leave the rails at 24v for now and just up the bias more and see how far I can go before heat challenges me. Then run some max output tests into 4 & 8 ohms, see how the square wave looks and also check THD.

Just one more question if I may. Is there a means to measure & see where the Q point is on each Mosfet load line using a scope or DMM and is there any benefit me doing this when ramping the bias to its maximum before heat issues arise? I know clipping will tell me a lot but I'd also like to see where the Q point is relative to power out and heat. This is purely an exercise to help me understand the F5 biasing.

I'm overseas at the moment and in ISO to get out of the country I'm in (bloody money grab). I've a new chassis waiting at home so will fit it out then follow up as stated. Will provide some feedback here once all done.


Many Thanks
 
Last edited:
There are a few things I can say with relative confidence, and there a few that I'll try to make sure that someone else either chimes in upon ... or recommend that you look up.

Your "order of concern" may be a bit off (to me). Here's a VERY general (and likely quite a bit wrong) walkthrough. I'll give it a bit of a go in very general terms with some super-broad (and likely incorrect) assumptions. I often say, the surest way to get the correct answer on the internet is to post the wrong one. :D

As mentioned, your JFETs (if Toshiba) will likely be just dandy with 32V rails. Linear Systems JFETs, maybe not. I can't remember where I read it, but something is tickling the back of my brain. If I find it, I'll share. Hopefully someone will chime in or you can find it as you read more about the F5T.

I am more interested in WHY you want to up the rail voltages. You still haven't fully answered the original question. You said you want more power.... but into what load? I assumed 2 ohm (but asked again for clarity).

I think I know where you picked a few pieces from the article.

"Increasing the power supply voltage is the obvious way to get more power out of an F5. You can simply raise the supply rails to +/-32 Volts and get 50 watts into 8 ohms right away without other modification"

Yes, if you up the rail voltage to 32V (lets assume that the F5 swings to the rails, but it does not) then you move from 48Vpp to 64Vpp, and yes, you will increase the maximum power output (into some load), but what's that load? More importantly, what will it do into the load that matters to you?

What effect does increasing the rails 33% (24V => 32V) have on idle heat dissipation assuming the Iq stays the same? In general, will this increase your max available power into a higher impedance (say 16ohm) load or a lower impedance (say 2 ohm) load?

What effect does increasing your bias current 33% (1A3 to 1A73) have on idle heat dissipation assuming the rails stay the same? In general, will this increase your max available power into a lower impedance (say 2ohm) load or a higher impedance (say 16 ohm load)?

Overall, what's the relationship of voltage and current to power => idle heat dissipation => heatsink temps for a Class A amp.

If a general rule says that my output devices are intended to only go up to ~35W max dissipation per device, and I want to run my rails at 32V, what's the highest current per device I should try (assuming each device sees the same current and sees the full rail voltage)? Same question for 24V rails.

If I need more total bias current (Iq) to reach my Class-A power target into a particular 'lower impedance' load, but I can't increase current per device, what's a potential solution?

Those are some of the things that I might think about in your situation. Admittedly, I may be thinking about it all wrong.... :D I'll learn alongside as someone with more direct knowledge chimes in. Hopefully soon. :rofl:

Edited to add - In all my typing; I forgot to answer your question about Q point. Yes, it can be done, but not practically with a standard scope or DMM (to my best knowledge). You'd need something a bit more sophisticated that can run the curves for you. There are a lot of great posts re: DIY type curve tracers. Also, sorry for the situation you're in. Hope you're back home soon.

Edited again to add - For the output devices on the F5 (not sure which ones you chose), the curves are likely out there somewhere already if you want them.
 
Last edited:

Allan

Member
2004-03-24 6:38 am
Perth
thanks AIMH.

my apologies for not providing earlier..
I am running the Toshiba 2SK170/2SJ74 6-8mA.
My current speakers are 4.7 min, nominal 6.8 ohm @ 87dB / watt / m
My PSU caps are 50V 20%

I'd like a bit more power simply for experimental reasons but also I do plan to go the F5T road at some stage, so this is also a stepping stone learning curve for me. I intend changing my speakers soon and don't want to be limited in choice because of speaker impedance / amp load ability, I prefer the amp remain load capable as the standard F5. If I do increase the rails to 32V I'd prefer to leave them at that if the amp pans out where heat isn't an issue

I hoping I'll get in the region of 50W @ 8 ohm, more into lower loads without it diminishing to much

On the current requirement side, I had planned going with a 600VA 24V transformer.. ~240V in Australia. I'm ok with removing the current limiter and did consider that but forgot to mention it.


Thanks Allan