Why 4 Ohm Loads Stress Your Amplifier

#1 is an Accuphase P-600.

 

Figure it's an expensive amp.

Looks like it has 7 pairs of those Sanken big transistors each channel and huge heatsink. It will drive 2ohm all day long.

 

Only top shelf gets the big SanKen (MT200) devices...
would love to get them both... one power ball....

 

My Luxman L-530 will drive 4 ohm loads all day long, the "internal" heat sinks seem to be there hottest when it is at idle, 130 degrees ( class A )
IMHO Chassis design is just as important as how much heat sink an amp has.

In the L-530 the big heat sinks and driver boards are isolated from the rest of the amp in a cooling tunnel, the hot air rising out the top draws cool air in the bottom past the sinks and the driver boards that are mounted on the sides of the sinks.
When I recapped this integrated I was surprised to see that after baking at 130 degrees all these years all caps on the driver boards checked ok for capacitance and showed no signs of heat stress.

 

No kiddin...

Great output stage

 

130 deg in C or F? If it is F, that's not that high. If it is C, that's dangerous.....for the amp. You sure it's a class A? It's rated at 120W @8ohm. The power dissipation for each channel calculated is about 260W per channel or 520W total of the whole amp!!!! That's very hot. It has only 4 output transistor, that's a lot of heat for 4 transistors.

I was looking at the schematic and read the spec. This amp has very low distortion. The newer amps mostly has higher distortion, how is the sound compare to those that are higher distortion? I am just curious to know as I designed and build a very low distortion amp, even lower and 0.003% THD.

 

Depending on ambient temperature, the average temperature of the heat sinks is 125 ~ 130 F, That may not sound that high but you can't keep your hand on it for more than a second or two.

The L-530 is rated @ 15 wpc pure class "A" 120 wpc class A/B.
Idle current stabilizes @ about 220ma per channel. The amp draws 340 watts @idle.
The L-530 is single stage, meaning it has NO PRE-AMP, the tone controls are in the dual feed back loops ( one for high frequency's one for low) The volume control is passive. S/N ratio is 110 db line and THD is 0.007
The 2sc3858 and 2sa1494 are robust. http://www.semicon.sanken-ele.co.jp/sk_content/2sa1494_ds_en.pdf

what I notice the most about this amp is not the low THD but the low noise floor, you can put your ear right up to the tweeters and you will hear nothing, not even the slightest hiss.

I powered many a party with the L-530 in the 80s, it had no problem at all driving my HPM 900's AND my friends Proximity's at the same time at rated power, all night long.

My only gripe with the L-530 is they eat main filter caps, they are famous for this.

 

I studied the schematic of L-530. I think it is a high bias class AB that has the first 15W in class A driving 8ohm. This means it has 7.5W in class A driving 4ohm.

From calculation backwards, idle current is about 1A. This is almost like my amp that I design and build. But I only use +/-42V rail instead of +/-63V like the L-530I use the big Krell 50W class A chassis that has huge heatsink on each side. Still it gets hot with my lower voltage. You amp must really cook!!! My chassis is this one http://www.ebay.com/itm/BIG-Aluminu...e=STRK:MEBIDX:IT&_trksid=p2057872.m2749.l2649 With 300mm X 200mm X 50mm huge heatsink on each side. Still it's hot. I don't even know how the L-530 can survive.

The design of L-530 is very low distortion. You ever compare to the high end amps like Bryston, Krell, Mark Levingston etc.? I wonder how the L-530 compare to the high end power amp. I am very interested in the amp as my design is ultra low THD, there are quite a bit similarity between my amp and this one. Main difference is the midle voltage amplify stage (VAS). I also use 9 pairs of output transistors instead of only two pairs in the L-530.

 

The 530 is using better output transistors... only need 9 pair of those if you plan on arc welding

 

Looking at the data sheet of Sanken 2SA1294 and 2SC3858, 200W 17A transistor. I use TTC5200 and TTA1943 150W 15A TO-264 transistors they are not exactly small. Sanken MT-200 can handle more power and current, but not even close to 2:1.

2 of the Sanken maybe like 3 of my transistors, not even close to 9. Also, my transistor is 30MHz, not 20MHz. That's one big advantage of my transistors.

Also, there is disadvantage dissipating a lot of power in one package, you create a hot spot on the heatsink. Better to spread the heat in different spot on the heatsink with smaller transistors.

 

2SA1216 is 40mhz if you want a lil more speed

 

I decided not to use Sanken because they all have very high Cob and low hfe. Cob put high load on the drivers and more capacitance at the output. I pick the TTC/TTA for their very low Cob. I could have use MJWL3281/1302 200W transistors which I already have a lot, in favor of the TTC/TTA.

 

2SB1648/2SD2561 is ahead in Cob and gain and 45mhz to boot

 

No it is not ahead at all if you check the TTC/TTA if you look at the datasheet. Also, as I said, at very very best, those big Sanken is equal to 1 1/2 of the regular transistors in dissipation, beta droop and current.

Another thing you have to keep in mind, bonding resistance and lead inductance. No matter how big the die of the transistor is, if you only have one lead coming out, the lead inductance is the same. You need more transistors to lower the lead inductance. Don't say lead inductance is not important if you are talking about very low load impedance circuit at the output.

More than anything else, it's not available in both Digikey and Mouser. Digikey min order of one is 1000 and in October.

I gave it a lot of thoughts on these Sanken before I decided that it's not as good. Yes, for those that use one or two pairs, it's better, not for a design that starts with many pairs. I actually thought about going the other way, using TO-220 small transistors and use 20 pairs. There are a lot of advantage using more pairs. There is no accident that most of the high end amps use many pairs of power transistors. There are solid theory very well written in the book by both Cordell and Self that more pairs lower crossover distortion. Double the output pairs almost half the crossover distortion. Unless you run pure class A, you have to live with crossover distortion.

That's the reason people use less pairs in class A amp. They just worry about beta droop, not anything else. For high end amps, nobody worry power dissipation of each transistors, it's the crossover distortion, the beta droop that we worry about. If you use 4 pairs or more, you cover the power dissipation issue of the transistors.

 

Well, the A list of TOTL amps use MT200 devices and yield the best results and are low impedance ready...

Amps including Technics SE-A3MK2 - Sony TA-N1, Onkyo M-510, H/K Citation XX, Yamaha MX-10000, Marantz PM10 PM14 (interesting copper frame) Rotel RHB 10, RB 1090, RB 1592, Parasound Halo A21, Accuphase P600 - Pro Amps Ramsa WP9220 & WP9440 -

It seems when the engineers get serious, these surface

 

I am talking about Krell, Pass Labs type of real high end amps. Accuphase use many many pairs of output transistors.

We all have different philosophy in designing. You look at what is important and what's not logically. I explained in detail the reason not to use very high power transistors and use less already. Those are solid reason backed up by theories and books with graphs to show. As I said and say again, power dissipation of the transistor are NEVER a problem, it's the heat sink, the thermal resistance from die to heatsink that govern the thermal limitation.

Increasing Cob directly affect the stability margin of the amp. Cob put capacitance directly at the output and as load to the driver transistors. AND before you say those big brands know what they are doing. I can tell you in my test where I hang a 1500pF cap right at the output connector of the amp and with no resistor load. The famous Nelson Pass design Nakamichi PA-7 burst into oscillation. Only 1500pF!!!! I design my amp to be able to drive up to 0.15uF at the connector with no resistor load. That's why so many amp cannot drive those high end high capacitance speaker cables. You have to guaranty the amp is stable with 1000pF, 3000pF, 6000pF, 0.01uF........ Just because you can be stable with one value does not mean anything that it will be stable with another value.

Those high end cables that use many smaller insulated cables "knit" together for close coupling, can have capacitance like 6000pF and is hung right on the output of the amp, that's the killer of many amps. You don't want to add more to it with Cob.

 

OK, my thoughts on this ancient subject.

I abused a marantz 2240 in my teens. I put 20000 uf caps in parallel with the main amp caps, and drove a pair of 8 ohm speakers in parallel with a pair of 4 ohm subs. That receiver still lives on with the original output transistors. The transformer ran cooler, the lights on the dial dimmed less, and it sounded much better than without the extra caps. The heatsinks definitely ran warmer, but a couple computer fans kept that in check.

My marantz 4400 has been running two 8 ohm sets of speakers for about 20 years in parallel. The manual states specifically 8 ohm speakers only when it's in 2 channel. I didn't know that because I never had a manual. Pretty sure the outputs are OK, but I don't know yet. It just recently died on me despite years of playing the music loud enough to drop the ac line voltage enough to dim the lights to the bass.

Right now I still have my two pairs of speakers on the a & b of my pioneer sx-1250. That receiver was used as a pa amp for my friend's band for a couple years, and severely abused. If you try to push the buttons for all three sets of speakers, it turns off one set, the light goes out on the front. Not sure how that works, but I'm pretty sure it puts a&b in parallel like most receivers.

Seems as long as your amp can drive two sets of 8 ohm speakers, AND the speaker selector puts them in parallel, you can drive a single set of 4 ohm speakers.

Also, from what I've read, current to the speakers is highest right above, and just below the woofer's resonant frequency. Which is what stresses the amp, there's a spike in impedance right at the resonant frequency. Big woofers in small boxes make this worse.

 

You'll be surprise how low the power output even when you crank the amp up loud. People don't realize how loud is 20W. Music has very high crest factor ( the peak amplitude divided by average amplitude). Occasionally you have spikes of high amplitude, but mostly the output stay at under 10W even if you play very loud.

People make high power amp to give more headroom so when the spike comes in, it won't clip the amp. So even if you think it's loud, the average power output is still very low. Use a scope and look at the output amplitude and you'll see.

Also people use higher power amps to prevent distortion from going up, distortion of amp goes up with output power. Distortion tends to make the sound "heavier", more body and stronger. In dancing and disco environment, you might even like it better. But if you listen alone at lower volume, if it has the same amount of distortion, it's tiring after listen for a while. You are really talking about apple and orange here.

Another thing is manufacture test with constant input signal to test power handling, that's very harsh on the amp. Music has high crest factor, it's more forgiving.