No, there is no spike if the output tube is conducting. The spike occurs when the output tube gets cutoff. Let's go back to the basic definition of inductance:
When I went to school in the 60's it was stated that inductance was the electrical property that opposes a CHANGE in current flow through that circuit.
From Wikipedia:
Note the word "(creates)", this is where the extra 1.5 KV comes from.
In normal SE amp operation with no signal applied the current through the primary of the OPT is not changing so the OPT is just a resistor equal to it's DCR.
When signal is applied the current through the primary of the OPT is varied by the output tube. This creates a changing magnetic field that creates a voltage across the OPT primary that is "equal to the product of its inductance and the time rate of change of the current." The voltage is changing at an audio rate, causing the voltage on the plate of the output tube to rise above the B+ value as the tube conducts less than its quiescent current during the trough of the audio signal. Through mutual inductance this voltage will also be induced into the OPT's secondary which causes a current to flow through the speaker's voice coil. Energy is being removed from the magnetic system to operate the speaker so the plate voltage will remain within reasonable bounds.
When the load is removed with signal applied the changing current through the OPT primary still causes a voltage to be created across the OPT's primary, and a voltage to be induced across the OPT's secondary. This voltage is "equal to the product of its inductance and the time rate of change of the current." As long as the tube remains in conduction the rate of current change is at an audio rate, and the induced voltage is reasonable and will tend to force some current through the output tube causing the voltage across the tube to rise, possibly above the 2 X B+ value accepted by many as normal.
When the output tube is driven to cutoff in the absense of a load the current through the OPT primary is INSTANTLY cut off. The "time rate of change of the current" is NEAR INFINITE. This causes the induced voltage to rise rapidly without bounds until a current can flow somewhere, or the magnetic energy in the OPT can be absorbed by system losses. If there is no path for current (no load, tube cutoff) the voltage will usually rise until an arc occurs somewhere. In a typical audio amp with common audio tubes and good insulation in the OPT, the arc will occur at the tube socket from pin 3 (plate) to pin 2 (heater). If the heater is grounded, the arc dissipates the energy, doesn't blow anything and ceases. Multiple occurrences can fry the tube socket. If the heater is elevated, the cap from heater to ground blows shorted, and the amp develops a hum. Sometimes the arc happens in the OPT killing it, or inside the output tube, maybe killing it.
You are building a monster amp that has all possible arc paths well insulated. You have a really big OPT capable of storing a lot of magnetic energy. There must be a non destructive path to release this energy in the event the amp ever sees clipping with no load. This why you want a GDT or spark gap across the OPT primary. It must handle 2 X the B+ voltage or more in normal operation, and be able to dissipate all the magnetic energy that can be stored in your OPT.
"I will never run my amp without a load." "My amp is so big that it will never see clipping." OK, now....Why does the guitar amp see 4 X the B+ voltage in "normal operation?" What is the impedance of an 8 ohm speaker? Why all these questions?????????
An "8 ohm" speaker is near 8 ohms at 1KHz, maybe. It has a published curve that will have a resonant peak where the impedance skyrockets. Depending on the enclosure it is possible for this peak to be as high as 100 ohms. Usually it's 20 to 30. My Yamaha NS-10M's have a 32 ohm peak at 85 Hz and an 18 ohm peak at 1500 Hz. These curves are measured with a single tone swept sine wave. Who listens to those?
Music is dynamic, multiple sounds are happening at the same time. A speaker is a moving mass attached to a coil of wire suspended in a strong magnetic field. This is the definition of a voltage generator.
The cone can be moving in one direction when a bass drum hit or a bass guitar slap tells it to instantaneously reverse its direction. At this instant the speaker will generate a counter EMF voltage to oppose that change causing its instantaneous impedance to be very low, lower than its DC resistance!
On the other hand the cone could be moving fast in response to a drum hit, when a sound impulse tells it to slow down just a bit. The instantaneous impedance of the speaker appears higher than its curve value at that frequency.
So, it's possible for an amp to be instantaneously over loaded, unloaded, or lightly loaded due to the dynamic nature of music, and speakers.
Ignoring the dynamics, what happens to a single tone played at or near one of the resonant peaks? The high impedance load can not remove all of the magnetic stored in the OPT so the voltage induced in the windings will increase.
I have measured clean sine waves of 65 V P-P across my Yamaha speakers with an SSE being fed by bass guitar. That would correspond to about 60 watts....from an SSE! WTF????? No, the SSE can't make 60 watts, the speaker is about 30 ohms and it provides some assistance due to the flying woofer cone. The 65 V P-P corresponds to a plate voltage swing of about 1300 volts. The B+ is only 450 volts, so yes it's possible to exceed 2 X the B+ and I haven't even hit clipping yet.
The pictures are taken from the scope screen with the scope set for 10 volts per division and wired directly across the speaker. One picture was taken at 10 mS per division, and the other at 100 mS. The OPT was a One Electron UBT-3 (3000 ohms) and the SSE had a KT88.
Guitar amps often run into hard clipping. This causes the output tubes to act like switches alternating between saturation and cutoff. There is often a brief period where neither tube is conducting due to bias shift and stored energy. If the amp is operating into a mismatched load that is near resonance, the plate voltage can spike to several KV.
It is the magnetic energy that is stored inside the OPT that causes the "spike" that we talk about, and this is the only real concern that we need to protect from in a real world amp.
Attachments
Yes, thank you Tubelab for the excellent explanation! You've no doubt answered a lot of folks' questions, asked and unasked. I will put a GDT across the OPT primary, 5kV or greater.
Speaking of "Monster Amps with Massive Transformers"...
I picked up my shipment from Monolith today. Check out the pics.
Each OPT weighs 62 lbs, each choke weighs 24 lbs (that's as much as my HV power transformer!). The chokes are rated at 35H at 180mA, 100Hz. They'll be L1 on the 833 B+, choke input.
A little more machining to do, then anodizing, then assembly.
Speaking of "Monster Amps with Massive Transformers"...
I picked up my shipment from Monolith today. Check out the pics.
Each OPT weighs 62 lbs, each choke weighs 24 lbs (that's as much as my HV power transformer!). The chokes are rated at 35H at 180mA, 100Hz. They'll be L1 on the 833 B+, choke input.
A little more machining to do, then anodizing, then assembly.
Attachments
Last edited:
ONLY 62lbs !!!!!!!!!!!!!! don't you feel cheated....joking aside I am envious. You can not beat plenty of iron...big trannies are like 12 inch woofers you need to move that Air and it can not be done with only 4 inches....size really does matter
man i'm wishing i could hear this beast when you are finished. your project is
as cool as cool gets.
so much inspired this has me start a parallel 300b mono block build as my first tube build.
this trans look like exactly the look i'm after.
please please keep the pics coming magz
and thank you for being an inspiring builder
So... You'll have roughly 110 lb of iron per channel. I hope you're planning to put casters on the chassis.
Or perhaps a big eye bolt at the center of gravity of the amp, and a ceiling mounted winch in your listening room...
~Tom
Actually a bit more. There are 6 chokes and 4 power trannies in addition to the OPT, per channel.
I plan to invite a few friends for a "listening party"...after they help me carry them out of the basement.
Dave from Landfall...if you're reading this, this is a true test of your aluminum chassis!
About 150 lbs per side is my estimate, give or take 5 lbs or so...and I saved weight by going two-stage, solid state rectification, and aluminum chassis.
awesome, what is the clearance between the HV and the secondary terminals?
the biggest traffo i ever made weighed in at about 22kgs...
