Why 4 Ohm Loads Stress Your Amplifier

Discussion in 'Solid State' started by 240 Volts, Feb 14, 2006.

  1. 240 Volts

    240 Volts Strange, yet oddly normal

    Messages:
    522
    Location:
    West Yorkshire, UK
    There have been quite a few questions lately regarding the safety (or otherwise) of driving 4 ohm speakers from amplifiers only rated to drive 8 ohm loads.

    Rather than just add to the mass of unsubstantiated conflicting opinions (e.g. "amplifier XXX is built like a tank, so it must be able to drive 4 ohm loads"), I thought I'd attempt to apply a bit of science to the subject. Hopefully this will make people aware of some of the issues involved in driving low impedances.


    In relation to ensuring reliable long-term operation, three of the most important parameters for a transistor are it's voltage, current, and power ratings. Exceed any of these three at your peril.

    Voltage : In any half decent amplifier design the output transistors will have a sufficient voltage rating to withstand all "normal" operating conditions. Only abnormal events such as lightening strikes are likely to cause failures.

    Current : The transistors need to be able to handle the maximum currents taken by the load. Speaker impedance can (and does) vary considerably with frequency, and often dips well below the nominal 4 or 8 ohm value. Amplifier designers are well aware of this fact and counter it either by using higher rated transistors with large peak current capability (good) and/or by incorporating current limiting circuitry (not quite so good, but better than blown outputs!). In practice a good amplifier design will withstand the ultimate over-current event - the accidental short circuit of it's outputs (but don't blame me if yours doesn't!).

    Power : Every transistor has a maximum power dissipation rating. This is not a fixed figure but varies with the temperature of the transistor. For example, a transistor might be rated to dissipate 100 watts at 25 degrees centigrade but only 20 watts at 100 degrees centigrade. The amplifier designer should provide cooling, in the form of heatsinks, to ensure that the maximum temperatures are not exceeded.


    It has been suggested by some here on AK that if you want to drive 4 ohm speakers with an amplifier only rated for 8 ohm loads, then everything will be OK if you limit the volume so that the output current is kept below that which would have been taken by the 8 ohm speakers. The following calculations will show just how wrong this is. :nono:

    I have used "Excel" to calculate the output transistor power dissipation for a typical 200 watt class AB amplifier when driving an 8 ohm load at full output and 4 ohm load at half output (so that the output currents are the same for both 8 and 4 ohm loads).

    The following assumptions have been made :
    (1) The load is purely resistive.
    (2) The output waveform is symmetrical about zero, so the calculation only needs to be done for the positive half cycle (180 degrees).
    (3) The amplifier output can swing to within 5 volts of the supply rails before clipping.
    (4) Bias current is neglected as it's contribution to the maximum power dissipation is relatively low for a typical class AB design.


    The basic amplifier topology for the analysis is shown below :


    [​IMG]


    The voltage, current, and power dissipation waveforms for 8 ohm load at full output and 4 ohm load at half output are shown below (NOTE: The "Output Transistor Dissipation" figure is the total power dissipation per channel, so to get a "per transistor" figure simply divide by the number of output transistors (not pairs of transistors) per channel).


    [​IMG]
    [​IMG]


    The dark blue trace is the DC supply voltage from the main filter capacitors.
    The sum of the output voltage (pink) and the voltage across the output transistor (green) must always add up to this DC supply voltage.
    The yellow trace is the current which flows through the output transistor and into the load.
    Output power (light blue) is simply the output current (yellow) multiplied by the output voltage (pink).
    Output transistor power dissipation (red) is likewise simply the transistor current (yellow) multiplied by the voltage across the transistor (green).


    So, for the 8 ohm load we have : Peak Current = 7.5 Amps, Output Power = 225 Watts, Output Transistor Dissipation = 85 Watts.

    And for the 4 ohm load we have : Peak Current = 7.5 Amps, Output Power = 112 Watts, Output Transistor Dissipation = 198 Watts.

    Despite keeping the maximum currents the same by halving the power into the 4 ohm load, the power dissipation in the amplifier output transistors is massively increased compared to full output into 8 ohms.

    The following graph shows how the power dissipation (i.e. internal heating) of this amplifier would vary with output power for 4 and 8 ohm loads. Note that the figures are "per channel" and so the heating is doubled for stereo operation.


    [​IMG]


    I have noticed that Pioneer specify output powers for their SX-x3x, SX-xx50, and SX-xx80 receivers into both 4 and 8 ohms, except for the SX-1280 and SX-1980 for which output power is quoted into 8 ohms only. Similarly, the SPEC 4 is rated at 150 watts into 8 ohms and 180 watts into 4 ohms, whereas the output power for the larger SPEC 2 is only quoted into 8 ohms (250 watts).


    [​IMG]


    My suspicion is that this is because the cooling on these higher power units is insufficient to allow them to develop the same (or greater) continuous power into 4 ohms as they can into 8 ohms (output power being measured by the old FTC regulations which required 1 hour pre-conditioning at 1/3 of rated output, i.e. near maximum heating of the output transistors).
    Pioneer could of course have specified an output power into 4 ohms for these models, but how good would it look if the output into 4 ohms was considerably less than that into 8 ohms?? Better for them simply not to give a power rating into 4 ohms. Of course, this is only my opinion and I'm open to suggestions from more knowledgeable AK'ers. :scratch2:


    To conclude then : For conventional class AB amplifiers the use of 4 ohm speakers greatly increases amplifier heating compared to when using 8 ohm loads - even if the maximum output currents are kept the same by exercising restraint with the volume control.


    OK guys, I've set myself up to be shot down in flames - who's going to take the first shot?? :D :D

    - Richard B.
     

    Attached Files:

    Last edited: May 22, 2006
    Jcrump likes this.
  2. bolly

    bolly Lunatic Member

    Messages:
    16,788
    Location:
    CFB Borden
    can you say that again all after Why? ;)
     
  3. jpdylon

    jpdylon non-active member...

    Messages:
    7,316
    Location:
    LaMesa (San Diego)
    I actually find this very imformative. It has confirmed my suspicions on why the amp gets so much hotter under a 4 ohm load.

    If there is a way to disipate the heat better (improved heatsinks, cooling fans, etc) would the transistors be safer, or does the wattage they disipate have the biggest effect on their lifespan?
     
    gunpoet likes this.
  4. ToTo Man

    ToTo Man NEVER ENOUGH TOTO!

    Messages:
    1,581
    Location:
    Scotland, UK
    Yeah I was wondering the same as Jordan, would placing a fan ABOVE the vent grille of the amp help reduce the -ve effects of a 4ohm load in terms of keeping it cool?
     
  5. phidauex

    phidauex High Technology Hippies

    Messages:
    447
    Location:
    St. Louis, MO
    Adding cooling definately helps, and I think lack of cooling is the biggest contributor to why some manufacturers don't spec 4 ohm or lower loads. I know that on my Phase Linear, the manual is very clear that you MUST use cooling fans on the heatsinks if you hope to achieve rated output through 4 or 2 ohms without toasting the transistors.

    If you could cryogenically cool your transistors, you'd be amazed at how much power they could dissipate. ;) Like the 266mhz freezer cooled 486 computers that people throw together sometimes.

    Thanks for the overview, I'm still absorbing all the math, but it seems to make sense, and definately correlates to my experience.

    peace,
    sam
     
  6. Nat

    Nat AK Subscriber Subscriber

    Messages:
    12,876
    The problem with fans is that if they are going to move air, they are probably going to make noise, so then, unless your amp is not in the same room as the speakers, its probable that you will turn the volume up. Probably you'll get ample cooling, but you are raising the noise floor.
     
  7. ToTo Man

    ToTo Man NEVER ENOUGH TOTO!

    Messages:
    1,581
    Location:
    Scotland, UK
    I think I'm gonna add fans to both my amps 'cos they even get hot after just after a few hours running 8ohm speakers at listening levels of 5watt max, so God knows how hot they'd get if I tried to really crank them on 4ohm speakers :yikes: .
     
  8. ToTo Man

    ToTo Man NEVER ENOUGH TOTO!

    Messages:
    1,581
    Location:
    Scotland, UK
    I'm pretty used to my noisey-as-hell liquid nitrogen fan in my PowerMac anyway so hopefully I won't notice the further increase in noise too much :D
     
  9. gyusher

    gyusher -don't go round hungry-

    Messages:
    8,537
    Location:
    Reynoldsburg, Ohio

    Use big fans turning real slow. . .silent. . .
     
  10. phidauex

    phidauex High Technology Hippies

    Messages:
    447
    Location:
    St. Louis, MO
    Here is a good site: http://www.silentpcreview.com/

    There are a lot of very quite fans available, and you can quiet down louder fans too.

    A good general tip is that a large slow fan is quieter than a small fast one, for the same given air flow. So choose the largest fan you can accomodate. If the air flow is too fast on the large fan, most of them can be run with a slightly lower supply voltage to slow them down (just don't go too low, or the fan can stall out). Two quality 12V, 5" box fans, running at 8 or 10 volts, will move a lot of air, and be very quiet. With a little care and vibration damping, they can be quieter than the ambient in your room.

    I appreciate a quiet room, I've got a cheap home theater PC in my room right now that I'm trying to quiet down... Just need to save up the money for a quieter power supply! The CPU fan is dead silent though, no noise heard outside the case.

    peace,
    sam

    EDIT: Gyusher posted at the same time as me! Good advice. ;)

    EDIT: Speaking of large fans moving slowly, check out http://www.bigassfans.com, the Big Ass Fans company. They sell HUGE fans, like 24' diameter! Their website is hilarious too.
     
    Last edited: Feb 14, 2006
  11. Surfacetension

    Surfacetension Well-Known Member

    Messages:
    857
    Location:
    Harmony NJ
    Maybe this is the problem I'm having with the Walsh -2 speakers I just got, I have isolated and am sure one speaker has issues OHM tells me that there is a protction circuit in these , Anyway as I turn the volume up they really don't get any louder and at 1/2 way they get staticy and the speaker circuit protect kicks in, I have a Pioneer SX 1010 and on the back of the amp it states 4 ohm's. The whole system is fine with the Pioneer speakers but the walsh -2's have issues
     
  12. glen65

    glen65 Member

    Messages:
    4,276
    Location:
    Richland county, Ohio

    There’s something you’re forgetting here. In most cases the receiver or amp is set up to be able to run 2 sets of speakers. This is provided that all four speakers are 8 ohm. When both speaker sets A and B are switched on they are simply connected in parallel with each other. This brings the total load to 4 ohms per channel. So you should be able to run only 1 set rated a 4 ohms.
     
    audioman00 likes this.
  13. sanyofreak

    sanyofreak Active Member

    Messages:
    390
    Location:
    Cleveland ,Ohio USA
    Volts,Amps,Watts, and HEAT !!!!

    240,
    Looks pretty good, should be an interesting read for many - this is the exact reason that I built my latest DIY amp with 3 Pairs of 250 watt rated output transistors(per channel), It can safely drive 4 ohm loads, but it is a 100WPC output at 8 ohms. Similar calculations and gasping at those numbers are what kept me to leave the voltage rails at about +/- 45 volts. I just could not come up with sinks big enough to go beyond what I did, and I didnt want a fan (not that they're bad), just did'nt want. I wanted safe 4 ohm operation however.
    I agree with You totally on why the really big amps, whether seperates or in monster receivers, dont rate into 4 ohms. (unless they are very expensive) And basically also true - if this derating at higher temps did not occur with BJT's, there would be no need to parallel output devices on the higher powered SS amps. Or if we could break the laws of physics and make a perfect heat sink (0 degree rise/W)
    Heat is the death of all electronics (basically), so to add a cooling fan to any unit that has to dissipate power is a good thing, You'll add to its longevity in most cases. Ideally, I'd want a fan below my equipment blowing air up through it, but any way of moving fresh air into the case (and warmed air out) will help. It's very surprising to observe the amount of heat rise above ambient that even a couple of watts of continuous dissipation can cause.
    This is a good read, and also reveals what a herculean feat it really is, to actually double an amps output into half the load, even with the overkill I designed into mine, it clips at around 175W/ch into 4 ohm loads. That also includes 1000VA of toroidal transformers, and literally hundreds of thousands of uF's of filter caps - on a 100W/ch amp.
    This is an area where the other classes of amplifiers (class G, class D, the Carver "tracking downconverter", etc.) have a thermal advantage over conventional AB designs. Well OK, I'm not so sure on class G, perhaps depends on the number of rails to get a substantial thermal advantage over AB. Thanks again 240, a good solid read as far as I can tell, and my experiences back You up. But I'm stickin with those heat wasting ineffecient AB amps, the well designed ones sound like nothing else.(oops --to my ears, that is)
     
    audioman00 likes this.
  14. Robisme

    Robisme Sansui Enthusiast Subscriber

    Messages:
    11,802
    Location:
    Martinez California
    If the amp is rated for 8 ohm's min, when using a and b you must use 16 ohm speakers. Some amps/receivers series (for the lack of a better term) the out puts. You can tell by switching your selector to a+b with speakers only connected to a. If you lose audio they are seriesed. If you do not, parallel.
    If rated for 4 ohm on a parallel a+b must be 8 ohm. =4 ohms.
    This has been said before.
    Rob
     
  15. Nakdoc

    Nakdoc nakamichi spoken here

    Messages:
    2,994
    Location:
    Nashville, TN
    Consider your bow shot. Your generous analysis has huge errors...sorry.
    An ideal amplifier, like the Levinson ML 2 (25w RMS), will deliver twice the current if the impedance is halved, and twice again if halved again. So this amp is 25W into 8 ohm, 50W into 4, 100W into 2, 200W into 1....etc. The ML2 is also a nice example because it is class A. The same current flows across the outputs with or without a load. This enables the amp to drive short circuits with no damage.
    Maximum amplifier heat is when running at 1/2 rated power. Run a 100w amp at 100w and it will warm up, but run it at 50 and HOT!. In the 100W example, the load is what gets hot! I won't explain the math of why the amp doesn't heat, except to point out that the specified test for amp power requires operation at 1/3 rated power until things heat up to max, and then the measurements are made.
    You have to do the math in peak AC terms, not DC or RMS. The picture changes.
    Secondary breakdown, not excessive peak current (and not peak voltage) are what shorts outputs. Secondary breakdown is often called the Safe Operating Area of a device. Designers use SOA analysis because almost all loads are reactive. The European DIN standard speaker load is reactive, not resistive, and contributes to the "lower power" that Euro stuff has, even though it plays louder than rated!
    The limit for most amps are fuses. A fuse is purely a current device, opened by heating, which is an RMS function. If an amplifier can deliver 5amps RMS through a 4 amp fuse, the fuse will open! At 4 amps we might get 50 watts peak at 8 ohms, but at 4 ohms we don't get twice that, because the fuse opens.

    I hope this plain English "long story made short" helps to clarify things.
     
  16. Kegger

    Kegger Anything can be S "MODed" Super Mod Subscriber

    Messages:
    12,226
    Location:
    Michigan,warren
    I agree a very well thought out and accurate account!
     
  17. wineslob

    wineslob Wiping out white van

    Messages:
    2,989
    Location:
    The West Coast
    Good article BTW! A bit generic, (but who would want to read a novel on this subject), helpful to most I would think. One other consideration and a cautionary warning. Power supplies. most of the power supplies in recievers are not designed for the amplifier to be over driven, A good example is the chart above. If an amplifier cannot double (or at least come close) its output into double the load ie: 4 ohms, IMO it's not a very good amp, ( 120w/ch 8 ohm 150 w/ch 4 ohm :thumbsdn: ) it may be a decent sounding amp, but some cost cutting was performed in its design, usually in the power supply/output transistors. This is one reason I stay away from recievers. Also to use fans to provide extra cooling for an overstressed amplifier is like burning a candle, sooner or later you will run out of wick. :nono:

    BTW I love this forum. We really have some good people here. I really enjoy discussions like this. Keep em comming! :thmbsp:
     
    Last edited: Feb 14, 2006
  18. Whitehall

    Whitehall Super Member

    Messages:
    2,199
    Location:
    Silicon Valley, CA
    The thermal heat load on the transistors and their heat sinks should be integrated over the cycle (RMS). Peak current (90 to 100% peak) only lasts a small portion of cycle.

    It would then follow:

    dissipated power = (I X I ) X R (corrected from "/" to "X")

    here R is the transistor's resistance and I is the RMS current in amps. Power transistors have a significant thermal mass so time averaging over a cycle is generally OK.

    Since power is V X I for pure resistive loads as assumed, for the same power at 8 ohms one would see twice the current at 4 ohms but produce four times the waste heat in the transistor.

    This doesn't jive with your curves and their summary boxes. Maybe you could further explain your methods.

    I will grant that semiconductors can suffer from internal hot spots that might be more limiting that total heat rejection to the heat sink. They can fail from local heating too.

    Now consider the case where the load is not purely resistive. Current and voltage cycles no longer overlap so one gets currents and their resistive heating without delivering real power. Ergo, for any delivered power (and hence sound level), more waste heat gets generated in the transistor.

    The problem is that low impedance speakers typically have increased deviations from pure resistive loads and so not only do they need more real power and its current depands, they need more imaginary power which adds yet more current and more heating.

    What's "Transistor Volts" and "Transistor power" in your curves?
     
    Last edited: Feb 14, 2006
  19. 240 Volts

    240 Volts Strange, yet oddly normal

    Messages:
    522
    Location:
    West Yorkshire, UK
    Nakdoc, I very deliberately stated that I was talking about class AB amplifiers - not class A which is a whole different ball game.

    Agreed that speakers are reactive loads which makes the transistor SOA an important consideration. I omitted to mention it as I didn't want to complicate matters too much. Of course, if the amplifier in my analysis uses MOSFET output devices then SOA doesn't come into play as MOSFETS don't suffer from secondary breakdown. :)

    - Richard B.
     
  20. 240 Volts

    240 Volts Strange, yet oddly normal

    Messages:
    522
    Location:
    West Yorkshire, UK
    All the power figures in my analysis are averaged over the cycle, I didn't call them "RMS power" for fear of upsetting those purists who (correctly) point out that there is no such thing as RMS power!!

    I should have made it clearer what "Transistor Volts" is on the curves I showed. It is the voltage across the transistor, i.e. Vc-e. This voltage multiplied by the "Output Current" through the transistor gives the instantaneous "Transistor Power" which is then integrated over the half cycle to give the "Transistor Dissipation" value.

    I can try to explain the details of my calculations more fully if you are still interested (or send you the Excel spreadsheet if you give me your e-mail address).

    - Richard B.
     
    Last edited: Feb 14, 2006
    merlynski likes this.

Share This Page