You can get small miners with a couple TH/s for very cheap-- anyone of your level of technical competence should have at least that.
If that would be the case, then signet would have at least a few TH/s. However, it is mined on CPUs instead, and there are no plans to change it. Also, other networks like Lightning Network are not mined either: mining is not for everyone, and never would be. If buying some coins directly is more profitable, than mining them, then a lot of users and developers won't have any miners beyond CPUs at all. Also, signet design shows it quite clearly, because if developers would want to see any ASICs on signet, then signing a new block template after checking every 2^32 hashes wouldn't be needed, and signed messages would be much more flexible than that.
If not to mine testnet you could play the solomine lottery.
I think both is good, but it would take some time and effort, to set it up properly:
For that reason, I thought about something different: what about using Merged Mining to produce testnet chain?
I guess it could be beneficial for many reasons, for example to see, if some non-standard transaction is consensus valid, and accepted by other nodes, or not.
mining support in the node software is in a sad state because of too much time with not enough of the technical community and almost no developers mining. That should change.
It wouldn't change, if people would keep rejecting Merged Mining as a bad idea. Most solo miners will never get a block in their lifetime. Which means, that they should get something else, while also mining Bitcoin. Only then they will have any reason to do so. In all other cases, it would be better for them, to simply buy some coins directly, and not mine anything, or mine altcoins, and sell them for BTCs. I can only guess, how many altcoins wouldn't exist at all, if mining would be better supported by the Bitcoin Core client, and if users could mine a bunch of satoshis over LN, or do other things like that.
If there nanoscale mining among node operators were more common it would also open up some useful options for improved sybil resistance and anti-DOS
For now, I have my own "nanoscale mining", where producing a double-spend for any transaction can be as easy or hard, as users want:
https://bitcointalk.org/index.php?topic=5551080.0And to test it properly, only CPUs are needed. Having more power is often not worth it, because another party will usually burn more electricity, to double-spend it, than it could earn, by taking coins, protected in that way. Also because Proof of Work is not the only protection, but it is complementary with other things like multisig, locktime, and so on.